discrete current mirrors

[Marte Schwarz]

Hi all,

there are much schematics around desribing how to design a current mirror on
a integrated circuit. But how to to this in good quality in discrete parts?
I want to built something like good old but obsolete OPA660.

in the Task "Low current source" Winfield published the following schematic:
.. 50nA current source without high-value resistors
..
.. 9V --+--- 13k ---+--- 1.0M ---,
.. | | |
.. e e gnd
.. b --+-- b
.. c | c MAT03
.. | | |
.. +-----' '----------O -->
.. | 50nA
.. 1.8M
.. |
.. gnd

I can not find a dealer for MAT03 and 7$ per piece seems me a little high,
isn't it. Do I need a npn-pair matched to the pnp-pair? would be the best
solution, but where to find.

Any ideas?

Marte

 

[Jim Thompson]

Hi all,

there are much schematics around desribing how to design a current mirror on
a integrated circuit. But how to to this in good quality in discrete parts?
I want to built something like good old but obsolete OPA660.

in the Task "Low current source" Winfield published the following schematic:
. 50nA current source without high-value resistors
.
. 9V --+--- 13k ---+--- 1.0M ---,
. | | |
. e e gnd
. b --+-- b
. c | c MAT03
. | | |
. +-----' '----------O -->
. | 50nA
. 1.8M
. |
. gnd

I can not find a dealer for MAT03 and 7$ per piece seems me a little high,
isn't it. Do I need a npn-pair matched to the pnp-pair? would be the best
solution, but where to find.

Any ideas?

Marte

There are a few additional bits of information needed...

Desired output current

Voltage range of the load

Absolute tolerance

Allowable variation of current as load voltage changes

Temperature variation

(There are several problems with the MAT03 design.)

...Jim Thompson

 

[Tim Wescott]

Hi all,

there are much schematics around desribing how to design a current mirror on
a integrated circuit. But how to to this in good quality in discrete parts?
I want to built something like good old but obsolete OPA660.

in the Task "Low current source" Winfield published the following schematic:
. 50nA current source without high-value resistors
.
. 9V --+--- 13k ---+--- 1.0M ---,
. | | |
. e e gnd
. b --+-- b
. c | c MAT03
. | | |
. +-----' '----------O -->
. | 50nA
. 1.8M
. |
. gnd

I can not find a dealer for MAT03 and 7$ per piece seems me a little high,
isn't it. Do I need a npn-pair matched to the pnp-pair? would be the best
solution, but where to find.

Any ideas?

Marte

Zetex sells discrete matched pairs already configured as current
mirrors. They're in the Digi-Key catalog.

 

[Walter Harley]

I can not find a dealer for MAT03 and 7$ per piece seems me a little high,
isn't it. Do I need a npn-pair matched to the pnp-pair? would be the best
solution, but where to find.

Digikey sells MAT03 (as well as other matched transistors, as Tim
mentioned).

The LM394 NPN matched pair is widely available.

I don't understand why you would need to find an npn pair matched to the pnp
pair. The matching is only important between the two transistors that make
up the mirror, because you are trying to ensure the same Ic versus Vbe
relationship.

Mouser sells some very inexpensive matched pairs, both NPN and PNP - just
search for "matched NPN", for instance. The ones I saw are not so good for
the low-current source that we were discussing, because they have much
higher collector leakage current than the MAT03.

Note that the design Win gave was for a dividing current mirror with an
adjustable, large, ratio. It is not what you would want for a normal (1:1)
current mirror.

 

[Pooh Bear]

Hi all,

there are much schematics around desribing how to design a current mirror on
a integrated circuit. But how to to this in good quality in discrete parts?
I want to built something like good old but obsolete OPA660.

in the Task "Low current source" Winfield published the following schematic:
. 50nA current source without high-value resistors
.
. 9V --+--- 13k ---+--- 1.0M ---,
. | | |
. e e gnd
. b --+-- b
. c | c MAT03
. | | |
. +-----' '----------O -->
. | 50nA
. 1.8M
. |
. gnd

I can not find a dealer for MAT03 and 7$ per piece seems me a little high,
isn't it. Do I need a npn-pair matched to the pnp-pair? would be the best
solution, but where to find.

Hmmm............. I can just about remember MATs. Micro alloy transistors. Very
very old.

I suggest you use something more recent !

Graham

 

[Jim Thompson]

Hmmm............. I can just about remember MATs. Micro alloy transistors. Very
very old.

I suggest you use something more recent !

Graham

Are you THAT OLD ?

MAT03's are relatively modern parts from Analog Devices.

...Jim Thompson

 

[Spehro Pefhany]

Hmmm............. I can just about remember MATs. Micro alloy transistors. Very
very old.

My guess is that the MAT is for "matched".

They're pretty decent, if you don't mind the price and single-sourced
aspects:

http://www.analog.com/UploadedFiles/Data_Sheets/410572853MAT02_e.pdf
http://www.analog.com/UploadedFiles/Data_Sheets/23265255MAT03_c.pdf

I had not noticed these ones previously:
http://www.analog.com/UploadedFiles/Data_Sheets/298405564SSM2210_c.pdf

Is there any reason why you want to use discrete components? An op-amp
will make a lot of the errors go away if you don't need to deal with
high frequencies.


Best regards,
Spehro Pefhany

 

[Walter Harley]

Hi all,

there are much schematics around desribing how to design a current mirror
on a integrated circuit. But how to to this in good quality in discrete
parts? I want to built something like good old but obsolete OPA660.

By the way, what's wrong with using the OPA860, which seems to be the modern
replacement for the OPA660? I don't know anything about OTAs, just
wondering.

 

[Pooh Bear]

Are you THAT OLD ?

I recall coming across them in a hobbyist article in one of the electronics mags -
probably back around 1969 IIRC.

MAT03's are relatively modern parts from Analog Devices.

That makes more sense - lol !

Graham