Can Diodes or JFETs Serve as Effective Voltage References?

[DragonPetter]

I am trying to find a 2.5V voltage reference, but the only part available in the library I HAVE to use is a 2.5V LDO regulator. I need to have the 2.5V reference supplied from a varying 10 to 16 V source.

What are other methods besides a zener or band-gap reference device that I can design a voltage reference with?

I was thinking of just using 4 0.6V drop diodes forward biased in series to get a 2.4V drop with a current limiting resistor. Is this feasible and why wouldn't I see this more often if it is?

 

[MATLABdude]

This is possible, but there are better ways of doing this (especially as diode voltage in forward active is fairly sensitive to current draw). One of the quicker and dirtier methods of making a voltage reference is with a Zener diode--much steeper I-V characteristic. If you put this in series with an appropriate resistance (consult the Zener datasheet) you should have something fairly stable over the range you indicated. There are more expensive voltage references (put out by the likes of Maxim IC) but many are just finer-tuned and more electrically stable Zeners.

 

[DragonPetter]

This is possible, but there are better ways of doing this (especially as diode voltage in forward active is fairly sensitive to current draw). One of the quicker and dirtier methods of making a voltage reference is with a Zener diode--much steeper I-V characteristic. If you put this in series with an appropriate resistance (consult the Zener datasheet) you should have something fairly stable over the range you indicated. There are more expensive voltage references (put out by the likes of Maxim IC) but many are just finer-tuned and more electrically stable Zeners.

Ya but the only zener I can use is a 5.6V and I need a 2.5V reference. I have no zeners available really. And those maxim ICs usually use a bandgap and I donæt have access to those either.

 

[uart]

How accurate and what range of current/temperature?

 

[Okefenokee]

You could use a common 5V zener diode and then split that voltage with two resistors set up as a divider. That's just a quick and dirty solution though. It's probably a waste of power.

They make voltage regulators that will suit your needs. http://www.datasheetcatalog.org/datasheet/nationalsemiconductor/DS009063.PDF". Look at the datasheet, under "typical applications", and you'll see that it's easy to implement.

I have a SPICE model in my library. You can add it to yours. I make no guarantees that the model is correct. I ran a quick simulation and saw that it works for your voltage inputs but the ripple is really high at the output unless the resistors are on the order of 10k..SUBCKT LM117HV 1 2 3
IADJ 1 4 48.4U
VREF 4 3 1.2782
RC 1 14 0.742
DBK 14 13 DLM117HV
CBC 13 15 2.479N
RBC 15 5 247
QP 13 5 2 QLM117HV
RB2 6 5 124
DSC 6 11 DLM117HV
ESC 11 2 POLY (2) (13,5) (6,5) 2.85
+ 0 0 0 -70.1M
DFB 6 12 DLM117HV
EFB 12 2 POLY(2) (13,5) (6,5) 3.92
+ -135M 0 1.21M -70.1M
RB1 7 6 1
EB 7 2 8 2 2.56
CPZ 10 2 0.796U
DPU 10 2 DLM117HV
RZ 8 10 .104
RP 9 8 100
EP 9 2 4 2 103.6
RI 2 4 100MEG
.MODEL QLM117HV NPN (IS=30F BF=50
+ VAF=14.27 NF=1.604)
.MODEL DLM117HV D (IS=30F N=1.604)
.ENDS LM117HV

 

[DragonPetter]

Accuracy is not that important as long as it is constant and temperature is just room temp as its a demo.

The other problem is that the original 2.5v zener only draws 20uA and this factors into a time constant that is used to drain a capacitor, so when I want to use the BZX84C5V6 that is provided in my library with an additional resistor divider, I have to draw about 5mA and this will throw off the capacitor discharge time horribly (its used in a bang-bang control, and the bangs will happen much faster if I go this route, and I don't want to worry about instability).

Basically I am trying to build this:
http://ludens.cl/Electron/solar2/solar2.html

with an EXTREMELY limited library provided for us, and no option to deviate from the cad library (it didn't even have FETs in it, but we got the maintainers of the library to atleast add a FET in for us).

I don't think the diodes in series will work either because when the input voltage changes the diodes drops will decrease too and you will have a race between the reference voltage dropping and the resistor divider into the + of the comparator dropping.

 

[DragonPetter]

I'm sorry, I don't think I was clear enough in saying that my limitation is this school parts library that cannot be added to so we really have our hands tied with parts we can use.

 

[uart]

You've got lots of options for such a simple circuit.

- You can use the 2.5 LDO regular that you have, I see no problem with that.

- You can use 4 forward biased signal diodes, that will also work (with perhaps a slight adjustment to R8, R10 to keep the switching thresholds much the same).

- You can use your 5.6 volt zener and a high impedance voltage divider.

- You can use the 5.6 volt zener "as it" and adjust R8,R10,R6 to keep the thresholds the same.

 

[DragonPetter]

You've got lots of options for such a simple circuit.

- You can use the 2.5 LDO regular that you have, I see no problem with that.

Well our LDO part will only go up to 12V input, so we may need to drop the voltage before we get to LDO input. It seems so silly to use an LDO regulator as a voltage reference. But you are right, this is a good option probably since we're so limited by parts.
Do LDOs require a minimum load with current draw to remain stable?

- You can use 4 forward biased signal diodes, that will also work (with perhaps a slight adjustment to R8, R10 to keep the switching thresholds much the same).

The problem I see with this is that as the battery/solar panel voltage decreases, the current through the 4 diodes will decrease and so their voltage drops will decrease which will result in a non constant voltage reference, and then the resistor divider voltage will be "racing" with the diode drops voltage reference.

- You can use your 5.6 volt zener and a high impedance voltage divider.
- You can use the 5.6 volt zener "as it" and adjust R8,R10,R6 to keep the thresholds the same.

Yes, but the 5.6V zener "BZX84C5V6" requires atleast a few mA to bias over the knee region, and then that will drastically change the current draw from the timing capacitor when the transistors shut off.

 

[skeptic2]

Another option is a JFET in a constant current configuration using the voltage across the resistor as your reference. JFETs can be biased so they are very stable over temperature.