You might as well use a 1% resistor - that will give a 1% accurate voltage at an exact specified test current!
- Vote on answer
- posted
16 years ago
-- John Devereux
1% zener diodes?
Loading thread data ...
- Vote on answer
- posted
16 years ago
Buried zener references are generally better than the alternatives such as band-gap references (well, except for little stuff like cost, noise and voltage/current requirements-- but for noise and stability they can be very, very good).
An LM317 is, at best, a "semi-precision" reference.
Best regards, Spehro Pefhany
-- "it\'s the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
- Vote on answer
- posted
16 years ago
On a sunny day (Wed, 25 Jul 2007 09:07:32 -0400) it happened Spehro Pefhany wrote in :
Zeners are noisy as hell, unstable, unreliable, expensive, inaccurate, bad current regulation, bad temp regulation, no way to adjust, have only 2 wires so they fall of easier then 3 or more wire parts ESPECIALLY WITH LEAD FREE SOLDER had to think of some more reasons you know.
It is orders of magnitude more stable then a 1% zener. LM317 is a bandgap reference, look up the temp specs.
I have heated up LM317 with soldering iron while monitoring on many digits Fluke, and it did not flip a digit. Now try that with the zener, also how to set a zener to 10.000V is intersting, let us know. With LM317 and 10 turn trimpot it is SIMPLE.
LM317 RULES
- Vote on answer
- posted
16 years ago
Known to us in the trade as a BandGap, so it behaves as a very low impedance, temperature compensated zener.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
- Vote on answer
- posted
16 years ago
parts ESPECIALLY WITH LEAD FREE SOLDER had to think of some more reasons you know.
See, for example, here:
formatting link
Yes, I know very well it's got an internal bandgap reference. So does a 78M05. The (at one time) common 1N829A reference zener has a 5ppm/K guaranteed maximum tempco. The LM317 is guaranteed to not change more than 1% (10,000ppm) over the temperature range, so 100ppm/K on the + side (using 125° C) and 125ppm/K on the - side (using -55°C). A standard 5~6V zener will have worse guaranteed tempco by serveral times (*not* several orders of magnitude), and could be compensated with a diode. We were able to get 100ppm reliably using this method with a cheap 400mW zener.
Fluke, and it did not flip a digit.
let us know.
Maybe the display is stuck on your Fluke. The voltage changes quite a bit with temperature-- it's a parabolic curve for a simple bandgap, which is what I believe it is (it's just a voltage regulator not a precision reference) Perhaps Jim can comment on the reference used in the '317? Schematic with parts values on p14 of the below-linked datasheet.
stuff on refs:
formatting link
LM317 datasheet:
formatting link
See page 6 lower left. An LM317 10V regulator will *typically* change about -120mV when heated to 125°C, or about -1.2% (so I guess there's some specsmanship with the 1% number!).
Best regards, Spehro Pefhany
-- "it\'s the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
- Vote on answer
- posted
16 years ago
Good buried zeners are way quieter than band gaps--e.g. the LM329, my personal fave. If you put a bunch of band gaps in series, you might get close. Any of those kinds of references rely on currents that have full shot noise, so eventually you have to spend power to get precision.
Somebody or other used to sell a voltage ref made out of a ruddy great MOS capacitor, pre-charged at the factory, integrated with a good quality closed-loop buffer. No current needed to flow, so there was no shot noise, and the cap discharged slowly enough to be quite a good reference.
Cheers,
Phil Hobbs
- Vote on answer
- posted
16 years ago
[snip]
That's the classic "Widlar" BandGap, except I designed such a circuit in the mid '60's, but didn't have the sense to publish :-(
The circuit has lots of resistor and device beta sensitivity.
Modern "BandGaps" built on CMOS processes do much better, due to current-mirror ratioing tricks... about 1mV over temperature.
Analysis of p14 is left as an exercise for the student ;-)
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
- Vote on answer
- posted
16 years ago
The LM399 is another. Ovenized so you can get +/-1ppm/K guaranteed (0.3ppm typ) with 7uV noise typical (vs. 200 times as much for an LM317).
I remember that.. sends shivers up my spine. Integrating soft errors. Yuk.
Best regards, Spehro Pefhany
-- "it\'s the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
- Vote on answer
- posted
16 years ago
On a sunny day (Wed, 25 Jul 2007 10:01:23 -0400) it happened Spehro Pefhany wrote in :
Fluke, and it did not flip a digit.
let us know.
Well, it has been some time ago, but I tested on 12V and on 24V IIRC and NOTHING like 120mV change, not even *1*mV change, maybe you should try too. It is sensitive to load variations, but in no way to temp variations. I do not remember who made the LM317, but I have tried this several times over the years with different ones. All that talk about zeners, those are only good for over voltage protection, and even then unreliable with a too high impedance. I have seen many defective zeners as fault source over the years in other equipment. I never had a LM317 fail, and I use these all the time.
- Vote on answer
- posted
16 years ago
Same here, except we (myself and the set) parted ways years ago, for the same reason! Thanks for confirming the 1%--that was my recollection.
Best, James Arthur
- Vote on answer
- posted
16 years ago
A linear supply with lots of expensive stuff connected and no crowbar is like crossing a canyon via rope with a safety hook. You still get there, usually ....
A sergeant at basic training accidentally did just that, forgot to engage his safety hook. He did reach the other side but upon discovering his mistake he nearly fainted and was literally shaking in his boots.
-- Regards, Joerg http://www.analogconsultants.com
- Vote on answer
- posted
16 years ago
Where'd you see the +/- 0.01mA? I couldn't find it.
It's admitted of course that the dynamic impedance of a zener isn't its highest attribute, but still, 10 ohms isn't a bad start, and it's easily improved with buffering.
Cheers, James Arthur
- Vote on answer
- posted
16 years ago
It's the test condition for figure 1 on the datasheet. You'd have to add another small error term if you deviate a bit from the specified current, and the tempco may no longer be 'guaranteed'.
Bootstrapping, as Jim T. mentioned, along with mentioning startup (the simple way by giving it a taste of unregulated current, or by doing something nonlinear so the regulation isn't degraded).
Best regards, Spehro Pefhany
-- "it\'s the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
- Vote on answer
- posted
16 years ago
It's not that bad--the zener's dynamic impedance is going to be much lower than the divider resistor's!
In the old days drift and accuracy were handled separately--first make something with a low tempco, then adjust it to the required accuracy.
In practice, +30v and two cascaded zener shunt regulators did pretty well, but you could also do something like this: (view in Courier font)
.-----------------------------. | | | | .-. | | | R1 | | | (trim for |\\ | '-' desired iz) | \\ | | | \\ | +----------------|+ \\ | | | \\-------o-------> Vout | | / | ,---' .--------|- / | / \\ ZD1 | | / | --- | | / .-. | | |/ | | R2 (trim for | | | | desired Vout) === | '-' GND | | '---------------------+ | .-. | |R3 | | '-' | | === GND
I think the above circuit's even self-starting if you choose an op-amp with sufficiently high bias currents and use low-enough valued feedback resistors. Otherwise, throw on a starter.
Cheers, James Arthur
- Vote on answer
- posted
16 years ago
Yes, yes, that's an old trick, I know it well. My objection was the 20mA sepcified current for the zeners - way to high! That's a non-starter in my book.
- Vote on answer
- posted
16 years ago
Ahh, thanks. Still, even a 0.5mA change in i(z) only results in a 5mV initial error; any affect on drift would be a 2nd order affair from there. So, the accuracy isn't fantastic, but the precision and drift are really pretty good.
Indeed. I've just posted such a circuit, only to find now you beat me to it!
Cheers, James Arthur
- Vote on answer
- posted
16 years ago
741 (and some brands of 709) OpAmp's will self-start in this configuration. I built hybrid circuits using this circuit between 1970 and 1973 at Dickson Electronics. Other OpAmp's may NOT necessarily self-start. So I handled that problem with a resistor/disconnect-diode scheme.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
- Vote on answer
- posted
16 years ago
My B-E junctions were integrated, 1mil X 1.5mil at that old time when device dimensions were huge ;-) Ran it at 100uA.
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave
- Vote on answer
- posted
16 years ago
Well, I was thinking of the LM324, what with its PNP inputs, not to mention fabulous precision & noise ;-), but there are others...
Oh, sheesh ... okay, you've shamed me into it:
"Bootstrapped reference with startup"
Vcc -+- | .-----------------------------. | | | | | | .-. .-. | | |R4 | | R1 | | | | | (trim for |\\ | '-' '-' desired iz) | \\ | | D1 | | \\ | V1-> +--|>|--o----------------|+ \\ | [*] | | | \\-------o-------> Vout .-. | | / | | |R5 ,---' .--------|- / | | | / \\ ZD1 | | / | | | --- | | / .-. '-' | | |/ | | R2 (trim for | | | | | desired Vout) === === | '-' GND GND | | '---------------------+ | .-. [*] V1 < V(ZD1) | |R3 | | '-' | | === GND
Cheers, James Arthur
- Vote on answer
- posted
16 years ago
There you go!
...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC\'s and Discrete Systems | manus |
| Phoenix, Arizona Voice:(480)460-2350 | |
| E-mail Address at Website Fax:(480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |
America: Land of the Free, Because of the Brave