[sdiy] Negative voltage references

Phil Macphail phil.macphail at liivatera.com
Mon May 21 19:45:37 CEST 2012



On 21/05/2012 02:50, "David G Dixon" <dixon at mail.ubc.ca> wrote:

>> David,
>> 
>> " I would say that this is not worth losing sleep over."
>> 
>> No offence here, but I'd argue on this one. I'd go with the
>> spec. I've been on several projects where the EEs didn't pay
>> attention to things like this and its bitten them. It doesn't
>> look good on their resume when there has been 10000 units
>> shipped and then most the units in manufacturing start
>> failing and the product line is shut down. Turns out it's the
>> original stock of parts has been exhausted and they've
>> ordered the next batch of parts from a different, cheaper
>> vendor. It suprises me that they can get a dozen or so units
>> from China to the states overnight and on the engineers desk
>> when this happens. More than once I've seen this happen.
>> 
>> Even though the parts that you have work, somewhere someone
>> will use a different brand or batch that will not exceed the
>> specs and they're hosed.
>> Also on common parts that are multi-sourced it's not a bad
>> idea to get a number of data sheets and go with the lowest
>> speced one because even though they have the same part number
>> they may have slightly different specs.
>
>I hear you.  However, in this case, I'm not even sure that this is a
>"spec"
>per se.  Nowhere in the datasheet does it say that one must pass at least
>1
>mA through the device.  the only reference to 1 mA is in the voltage
>output
>specs in the table, where the current is specified between 1 and 100 mA.
>It
>doesn't say that the device won't work properly if it is not passing at
>least 1 mA, and there is one plot where the curve is drawn at 0 mA output.
>Why would they bother plotting data at this current if this were not a
>viable current for operating the device?
>
>Also, in my case, I only use 79L05 in "non-critical" situations, where I
>don't really care all that much how close the reference is, or whether
>it's
>constant with temperature, etc.  In critical situations, I would use a
>bandgap reference.
>
>_______________________________________________
>Synth-diy mailing list
>Synth-diy at dropmix.xs4all.nl
>http://dropmix.xs4all.nl/mailman/listinfo/synth-diy


I agree with Jay here, 3-terminal regulators are easy to use, but this
ease can breed complacency. The 78xx parts are better than most, but there
are two general problems associated with incorrect loading on a regulator -

1. The bandwidth of the regulator increases as the load is reduced, so
there is a chance that the regulator can oscillate. This is more of a
problem with low-dropout regulators which don't have the emitter-follower
structure of the78xx, but they are more sensitive to point 2.

2. The regulator has a huge DC gain to ensure good load regulation, but
the consequence of this is that it can be very difficult to turn-off the
output device when the load is small. For the 78xx devices this means
removing the base drive to a darlington transistor pair. Without drawing
enough current from the transistors the output voltage will rise to 2*Vbe
below the supply. Hopefully the internal resistor feedback network will
provide sufficient load, but this current will have large process
variation, and even bigger temperature sensitivity. Hence the comments
from Neil and Jay. Most of the time things will be fine, but it is simple
to make it fine all of the time,


Phil.





More information about the Synth-diy mailing list