dB and hfe

[perfectbite]

Can dB and hfe be rendered as similar or interrelated measurements under differing conditions or are they entirely distinct from each other? Thanks.<p>[ July 03, 2004: Message edited by: perfectbite ]</p>

[Chris Smith]

Decibels is a out put factor, not a gain factor.

[perfectbite]

So, in looking at the gain for a transistor hfe is used but in looking at the output signal for an op-amp dB would be used regardless of the strength of the input signal? Thank you Chris.<p>[ July 03, 2004: Message edited by: perfectbite ]</p>

[dyarker]

Yes, but not usually. hfe is current gain of just the transistor. Often dB is used to reference the gain of a circuit, the transistor, it's associated resistors capacitors inductors, and the impedance of the load.<p>A transistor with a minimum hfe of 100 may have an actual hfe between 100 and 300. Also, the hfe decreases as frequency increases; and hfe increases as temperature increases. All of these are bad for designing a predictable circuit that can be repeated. By adding a resistor in the emitter path, the circuit gain is about the ratio of the collecter and emitter resistors as long as hfe is greater. With a 10K Ohm collecter resistor and 500 Ohm emitter resistor, the output voltage swing at the collecter/resistor connection will be 20 times the input voltage swing at the base.<p>volts dB = 10 * Log (10000 / 500) = 13dB (if I remembered that right. And, power dB is 20Log of the ratio.)<p>If, for example, you need 20dB gain for your project to work; you can change resistors or add a 7dB stage.<p>So yes, a "barefoot" transistor with hfe of 100 can be expressed as 10Log100 = 20dB gain, why bother?

[dyarker]

dB is a logrithmic ratio. The ratio of input to output is gain (ot loss). Signal to noise ratio. Etc.<p>Absolute measurements have a qualifier. Like dBm which stands for decibelmilliWatt. 0 dBm = 1mW. A signal of 0.5mW is -3dBm; and 6dB (no "m") weaker relative to a 2mW signal.<p>Op amps can have open loop gains (Av) over a million (60dB). Again, why? What you want to know is circuit gain in dB with the input and feedback impedances included.

[Dean Huster]

No two transistors, even those with the same type number (e.g., 2N3904), will have the same hfe. hfe varies also with collector current, temperature, etc. That's why amps are designed such that the stage gain is dependent upon the EXTERNAL components and not the transistor itself. If hfe (the ac version of the dc beta) had a large effect on stage gain, you'd have to specially select each transistor and repair would be a nightmare.<p>By the way, if an hfe of 250 implies that a change of 1µa on the base causes a 250µa change at the collector, I don't know why you couldn't express the gain in dB .... in this case, 48dB. But they just don't do that. Besides, since hfe and beta are such sloppy characteristics, there isn't much point in playing with other methods of expressing that figure.<p>I have seen circuits from legitimate biggie manufacturers that do select the transistors used in some special applications for a minimum hfe or some such. I don't remember the application, but it may have been in circuits with such high gain that the hfe did make a difference.<p>Dean

[perfectbite]

Dale. Many, many, many moons ago when I attended technical college in a very low level course of study I asked the slide rule teacher what the Log Log scale on my, given as a gift, case and all, Pickett slide rule was for. "Who knows?" was the answer. "Just concentrate on the front bottom scales." so it's back to Logs for me. Your comment about a 'bare footed' transistor helped me resolve my question(s). Thank You.

[perfectbite]

Thank you also Dean.

[Ron H]

<blockquote><font size="1" face="Verdana, Helvetica, sans-serif">quote:</font><hr>Originally posted by Dale Y:
Yes, but not usually. hfe is current gain of just the transistor. Often dB is used to reference the gain of a circuit, the transistor, it's associated resistors capacitors inductors, and the impedance of the load.<p>A transistor with a minimum hfe of 100 may have an actual hfe between 100 and 300. Also, the hfe decreases as frequency increases; and hfe increases as temperature increases. All of these are bad for designing a predictable circuit that can be repeated. By adding a resistor in the emitter path, the circuit gain is about the ratio of the collecter and emitter resistors as long as hfe is greater. With a 10K Ohm collecter resistor and 500 Ohm emitter resistor, the output voltage swing at the collecter/resistor connection will be 20 times the input voltage swing at the base.<p>volts dB = 10 * Log (10000 / 500) = 13dB (if I remembered that right. And, power dB is 20Log of the ratio.)<p>If, for example, you need 20dB gain for your project to work; you can change resistors or add a 7dB stage.<p>So yes, a "barefoot" transistor with hfe of 100 can be expressed as 10Log100 = 20dB gain, why bother?<hr></blockquote>
Dale, you got power and voltage dB's reversed. A voltage gain of 100 is 40dB. See
this site and this one.

[Dean Huster]

Ron, voltage and current ratios get the 20log(Vout/Vin) and 20log(Iout/Iin) respectively while power ratios get the 10log(Pout/Pin). You need to get out of the DRAM department more often! Take a well-earned vacation down here to SE Missouri and I'll treat you to some back-woods hick bluegrass (the kind you listen to, not the kind you smoke).<p>Seems to me I always used the LL scales on the slipstick to raise weird numbers to weird powers. Before the scientific calculator and it's cute little "Y to the X" key, the only way you could do that was to use a loglog slide rule or spend a bit of time with a logarithm table (courtesy of the CRC Math Handbook) and a bunch of interpolation to beef up the precision of the answer.<p>I still have the CRC Math Handbook I bought for my college classes back in the 60s .... do you, Ron?<p>Dean<p>(For those not in the know, "CRC" stands for "Chemical Rubber Co." and was the publisher of several math and scientific handbooks.)

[dyarker]

Thanx RonH. That's why I put the "(if I remembered that right ... )" in there. Its been many years since I've had to calculate to convert between linear gain, power dB and volts dB. At work its strictly dBm box-to-box; bridged (high Z) mode for working channels, term mode with correct impedence on the bench.<p>Maybe type out a memory crutch, I'll remember. If you would multiply the "regular" numbers, then add the logs of the numbers. If you would square, cube, etc a regular number, then multiply the log by the power.<p>When dB has a qualifier to make it a power unit: If the power is doubled, that's times 2 before taking the log (right side of formula). About 3dB. If the voltage is doubled, the current is also doubled; so the power is squared. That's times 2 on the left side(about 6dB). Therefore, volts to dB gets the 20 (10 * 2)and power to dB gets the 10. Without the 10 the result would be Bels instead of deciBels.<p>db = 10 * Log(Output Power / Input Power)
db = 20 * Log(Output Volts / Input Volts)
If input is greater than output (loss), dB is negative.<p>------------------------------------------
I haven't used a slide rule much since college (1971), but I had to use more of the scales and still have the aluminum Picket.<p>Want me to work up some comments?<p>Its kind of a shame slide rules were mostly killed by calculators. For most electronic work 2 1/2 or 3 digits of accuracy is enough (1% or 5% resistors, 5% or 10% capacitors). A proficient slide rule user can have the answer while a proficient calculator user is still tapping buttons.<p>[dang! typing at the same time as Dean again, and he types faster.]<p>[ammended per David Bridgen's correction in italics. Thanx Dave. Now we know why proofreaders exist.]
C U L -<p>[ July 06, 2004: Message edited by: Dale Y ]</p>

[bridgen]

Dale Y, <p>you say "dB is a power unit".<p>Not without qualification it isn't. <p>Without qualification it is just a ratio, as your correctly stated earlier. <p>My notes about dB notation might interest other readers. They're <p>here

[Ron H]

Dean, I think I got that right. I might take you up on that bluegrass offer though, even if you do send the dB cops after me.
I do still have my CRC handbook - the copyright is 1959. I still use it occasionally, generally for trig identities, integrals, and differentials. The log and trig tables in the front have been superseded by my calculator.<p>Ron

[Dean Huster]

I still have both of my Post rules, the 10" and the 5" (a cute little "mini-me" of the 10") and still use them when I can't find a calculator, which isn't that often. The trouble with slide rules was that it was a bear to keep up with where the decimal point was going to end up (except for the log-log scales!) unless you were highly trained and well-rehearsed in some of the methods for doing so.<p>It's interesting that not only did the calculator replace the tables, but it did it with typically twice the precision. I understand that the original HP-35 had some anomolies in their transcendental algorithms that made the last few digits of some of the results questionable, but not enough that it ever affected anything that most of use ever did. And that had to have been found by hp themselves because I know that I had never seen trig tables with the precision to check the calculator results. Maybe someone worked out a few common angles using standard math and trig ratios to find the error.<p>Pickett was neat in they had lots of specialty slide rules for things like electronics. But if you used them long enough, they made your fingers black and most of them ended up with the yellow paint chipping off. I always liked the bamboo rules best. And no, they didn't give you splinters!<p>Dean