This is a discrete Volt Reference
with bandgap for temperature compensation & stability.
The circuit is the most basic bandgap circuit by the schoolbook.
The output according to simulation is 1.292 Volt
It will stay very constant for temp 15-65 degrees Celsius.
I would like to know how well this will work.
Meself, to test temp stability of circuits use a normal lamp.
A 60 Watt light bulb will radiate heat. (As much as Class-A 60 Watt amplifier)
If you put a warm lamp close to your circuit,
you can then measure what happens .....
If anyone does a test, let me know results
Also, post if you have some other discrete bandgap circuit
Lineup
with bandgap for temperature compensation & stability.
The circuit is the most basic bandgap circuit by the schoolbook.
The output according to simulation is 1.292 Volt
It will stay very constant for temp 15-65 degrees Celsius.
I would like to know how well this will work.
Meself, to test temp stability of circuits use a normal lamp.
A 60 Watt light bulb will radiate heat. (As much as Class-A 60 Watt amplifier)
If you put a warm lamp close to your circuit,
you can then measure what happens .....
If anyone does a test, let me know results Also, post if you have some other discrete bandgap circuit Lineup
Attachments
lineup,
Bandgap references were developed for use in ICs where it is easy to get well matched devices. (Bandgaps depend on good matching of the transistors.) I don't think discrete bandgaps will perform very well unless the transistors are carefully selected and, even then, I wonder.
Note that in a simulation all instances of a particular device will be identical, and this will show unrealistic performance. You really need to use devices with slightly different characteristics to determine the effect of the differences.
Rick
Bandgap references were developed for use in ICs where it is easy to get well matched devices. (Bandgaps depend on good matching of the transistors.) I don't think discrete bandgaps will perform very well unless the transistors are carefully selected and, even then, I wonder.
Note that in a simulation all instances of a particular device will be identical, and this will show unrealistic performance. You really need to use devices with slightly different characteristics to determine the effect of the differences.
Rick
I am not sure why you are using a desecrate Band-gap ckt. This has lots of discrete components which will affect the output.
As a matter of fact, you can use the technique for testing this circuit.
Now let me come to the problems.
1. you will not get the same performance when you make the same circuit next time as he resistor's and BJT's Beta tolerance.
2. the temperature behavior will vary with the components. (doesn’t matter how close you keep them).
As a matter of fact, you can use the technique for testing this circuit.
Now let me come to the problems.
1. you will not get the same performance when you make the same circuit next time as he resistor's and BJT's Beta tolerance.
2. the temperature behavior will vary with the components. (doesn’t matter how close you keep them).
You'll never beat IC references such as the LM236, LM285, LM4040 or TL431, both in terms of price and performance. You'll need transistors on the same substrate, with accurately scaled dimensions. You can get identical transistors on a single substrate, but they're expensive and much harder to obtain than today's IC "zeners" (which derive their zener voltage from an internal bandgap). As an exercise it is nice though.
sawreyrw said:
yes, of course i can not argue, Rick
besides, there is not much point in building your own discrete bandgap
with or without use of monolitic transistor chips,
as a good and factory trimmed 1.25 Volt reference IC with low temp coefficient
does not cost us a fortune
it is more for the fun of it
to see what one can do & if you can get it work well
diyAudio
projects by the fanatics, for the fanatics
thanks, redrabbit
Was a good pdf - I downloaded it.
Another wellknown and great overview of different bandgaps
is this article by Bob Pease:
The Design of Band-Gap Reference Circuits: Trials and Tribulations
http://www.national.com/rap/Application/0,1570,24,00.html
Normally a bandgap has got a peak temperature, where v-ref is the highest.
We can adjust the circuit to set this peak at a certain temperature.
Say at +40 C.
We will have a slight voltage sink on both sides from this 'center temp'.
At +20 C it will be maybe 1.2495 V
1.2500 Volt at +40 = the peak.
And at +60 also 1.2495
This schematic in Bob Pease article
is showing a way to make the temp curve more flat in a bandgap.
Was a good pdf - I downloaded it.
Another wellknown and great overview of different bandgaps
is this article by Bob Pease:
The Design of Band-Gap Reference Circuits: Trials and Tribulations
http://www.national.com/rap/Application/0,1570,24,00.html
Normally a bandgap has got a peak temperature, where v-ref is the highest.
We can adjust the circuit to set this peak at a certain temperature.
Say at +40 C.
We will have a slight voltage sink on both sides from this 'center temp'.
At +20 C it will be maybe 1.2495 V
1.2500 Volt at +40 = the peak.
And at +60 also 1.2495
This schematic in Bob Pease article
is showing a way to make the temp curve more flat in a bandgap.
An externally hosted image should be here but it was not working when we last tested it.
Here is such a typical curve of bandgap.
My simulation of my discrete bandgap references
shows exactly this behavior and curve for voltage deviations.
Note: This sloop is at a very low level:
I have set up some circuits that only sinks 0.01% per +-20 degrees.
That is 1/10000 deviation from the peak of curve ( at 1.25 Volt ).
Typical curve:
My simulation of my discrete bandgap references
shows exactly this behavior and curve for voltage deviations.
Note: This sloop is at a very low level:
I have set up some circuits that only sinks 0.01% per +-20 degrees.
That is 1/10000 deviation from the peak of curve ( at 1.25 Volt ).
Typical curve:
An externally hosted image should be here but it was not working when we last tested it.
Anyone in their right mind would just buy one of the many great references available today. For some unusual stuff, check out Thaler. Not being of sound mind, and not caring about low voltages, I've built various discrete references using the 1N82x and similar zeners, some in ovens, some not. IMO, the best circuit buffers the zener with a good op-amp, and also sets it's own current. Put it in a small box with a light bulb and adjust the zener current until the tempco is near zero. You'll find the tempco of the gain resistors becomes a significant factor, but it's possible to get everything nulled out. What's really difficult is getting the overall tempco near zero *and* getting the output trimmed to an exact value.
Cool Thaler Stuff- refs, converters, amps
Cool Thaler Stuff- refs, converters, amps
You can rely on different semiconductor materials rather than current densities for temperature compensation, see this application example:
A PSU controller on a shoestring.
Most LEDs have a tempco close to -2mV/°C, and you can even find a LED+transistor combo in a single package: an optocoupler.
If you use an optocoupler in the regulator above, it may have "interesting" results, because the LED will add feedback to the error-amplifier transistor.
Some fancy couplers like the 6N139 have a red LED rather than a IR one, and the two transistors are separately accessible, meaning a higher ΔVf and no parasitic feedback.
The ΔVf for an identical current density is ~730mV, and remains very stable against temperature.
Perhaps usable if you want to brew your own micropower reference?
A PSU controller on a shoestring.
Most LEDs have a tempco close to -2mV/°C, and you can even find a LED+transistor combo in a single package: an optocoupler.
If you use an optocoupler in the regulator above, it may have "interesting" results, because the LED will add feedback to the error-amplifier transistor.
Some fancy couplers like the 6N139 have a red LED rather than a IR one, and the two transistors are separately accessible, meaning a higher ΔVf and no parasitic feedback.
The ΔVf for an identical current density is ~730mV, and remains very stable against temperature.
Perhaps usable if you want to brew your own micropower reference?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.