Back from holiday .... 3000+ photos later ... :-) I'll leave all the quoted material in as its useful: The question was ~ : Suggest why am I using a forced beta of < 1 with base and collector supplies powered from the same voltage so there is no current, power or voltage gain: >>Spehro Pefhany very very very correctly > wrote: >> > One obvious (?) purpose would be to have very low Vce (perhaps in >> > the >> > single or low-double-digit mV range), and to avoid the need for a >> > low >> > Rds(on) small-signal p-channel MOSFET. >>But that's pretty much true as long as the transistor is saturated. >>You >>don't get much more by oversaturating the transistor. > Errr... I don't think so! > Quick test with a jellybean TO-92 PNP transistor shows a 3:1 > improvement > going from forced beta of 5 to forced beta of 0.5 (11mV to 3.2mV) @ > 0.5mA IC. > That's a very respectable 6 ohms equivalent Rds(on)- similar to a > much > more expensive BSS84, and using a 8550 wot is so common you can find > them > stuck into the bottom of your shoe from a short walk around Taipei > or Shenzhen. Around here it's BC327s and their preferred gravitating place is the washing machine. Target location for final; product is Taichung, 2nd largest ROC city after Taipei. > Also it will not be as temperature-sensitive. At 50 forced beta > there's a whopping 57mV of drop, (still well saturated of course). Give the man a virtual Pepsi / Dr Peppers / whatever ... . 100% correct. The application is switching the divider chain from a 12v SLA battery (deja vu) to an AVR ADC. Higher enough values of resistor chain to begin to affect the accuracy * still impact the battery life adversely. eg 1 megohm at 12 volts is ~~ 12 uA or ~0.1 amp-hour in a year. A more realistic and almost too high 100k gives 1 AH in a year - far too high. One could use a much higher divider value and an opamp. It's a cost / space / other trade-off between using a high side switch to turn off the divider or adding an amplifier IC. In this case the divider won. Using a high side P Channel FET is the obvious choice and the circuit will support that if it proves to be a hassle free and cost advantageous decision. BUT there are reasons why this may not be so. Using a PNP high side switch works BUT for "normal" Beta levels in the say 10-100 range the switch saturation is significant wrt the measured voltage. As Spehro notes, it can be 10's of mV. BUT using a Beta of 0.5 (Ic ~= 0.5 mA, Ib ~= 1 mA) drops the saturation voltage to under 10 mV. As base current is readily available when the circuit is running I am happy to use the 1 mA. I get about 5 mV 'saturation'. When measuring 12 volts 10 mV = 1/1200 ~= < 1 bit error in 10 bits. As noted in the SLA discussion, a 0.1 volt error would be excessive (1/120 or about 7 bits). (10:1 = 3.32 BITS FWIW). Actual improvement will be somewhere in between depending on transistor, actual beta etc). My first attempt used a single floating transistor but this (stupidly) included the base drive in the lower divider resistor. Usually that's reasonably irrelevant. Here it's fatal. So I now have two transistors in a classic bipolar high side drive. A factor in the switch/amplifier decision is the elimination of 12v "sneak paths" when the system is off. These have been the source of problems in other instances and the switched divider based system eliminates these. Use of an op-amp would allow small but potentially [pun only vaguely intended] off currents into the opamp and thereby into the processor and thereby ... 'all for the want of a horse-shoe nail' (Google knows). Also, if an eg smaller capacity alkaline battery is used the whole system off current will be zero (<< 1 uA) and off battery life will be years. * Increasing input impedance first affects the dynamic response due to time constant effects and these are not overly crucial here, but still higher values impact accuracy due to bias currents. Russell McMahon -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist