Favourite circuits for CC generator ?

[ChrisGibboGibson]

I want a constant current generator that will produce 1 to 2 mA over an output
range of 2 to 6 volts running from a 10 to 15 volt supply.

It seems quite an easy task but here are the full requirements.

The exact tolerance of the current isn't that critical. 1mA would be fine,
1.2mA would be fine 1.6mA would be fine. Whatever is settled on, the tolerance
between different units could be as high as 10% without it causing any
problems.

Drift over time, in the long term, or with temperature isn't a problem either.
Even wild drift of up to 20% can be tolerated.

The difficult part is that over a 20mS period, and with the load varying such
that the CC output will swing from 2 to 6 volts, the current must not vary by
more than 0.05%

I'm up against a wall trying to come up with a circuit that meets this spec. No
doubt one of you lot knows one off the top of your head.

Gibbo

 

[John Larkin]

I want a constant current generator that will produce 1 to 2 mA over an output
range of 2 to 6 volts running from a 10 to 15 volt supply.

It seems quite an easy task but here are the full requirements.

The exact tolerance of the current isn't that critical. 1mA would be fine,
1.2mA would be fine 1.6mA would be fine. Whatever is settled on, the tolerance
between different units could be as high as 10% without it causing any
problems.

Drift over time, in the long term, or with temperature isn't a problem either.
Even wild drift of up to 20% can be tolerated.

The difficult part is that over a 20mS period, and with the load varying such
that the CC output will swing from 2 to 6 volts, the current must not vary by
more than 0.05%

I'm up against a wall trying to come up with a circuit that meets this spec. No
doubt one of you lot knows one off the top of your head.

Gibbo

Small mosfet, source-lead current sampling resistor, bandgap voltage
ref, opamp closing the loop. I could post a schematic to abse if you
want.

Oh, source or sink?

John

 

[ChrisGibboGibson]

Small mosfet, source-lead current sampling resistor, bandgap voltage
ref, opamp closing the loop. I could post a schematic to abse if you
want.

Oh, source or sink?

Source.

Will that meet the spec ?

Simulations say not though I haven't tried it in real life.

Gibbo

 

[John Larkin]

Source.

Will that meet the spec ?

By a couple orders of magnitude!

Simulations say not though I haven't tried it in real life.

What do the sims show? Where is the error?

John

 

[John Woodgate]

I read in sci.electronics.design that ChrisGibboGibson
m01.aol.com>) about 'Favourite circuits for CC generator ?', on Thu, 30
Sep 2004:

The difficult part is that over a 20mS period, and with the load varying
such that the CC output will swing from 2 to 6 volts, the current must
not vary by more than 0.05%

I think you need an output source impedance of 10 Mohms or more (taking
6 V/1 mA and 2.001 V/1.0005 mA to make the numbers easier). That isn't
impossible, IIRC. One of my colleagues designed something like that for
a test instrument but it was a long time ago.

But I'm not suggesting a 10 kV supply and a 10 Mohm resistor. The source
was a particular type of transistor that had a very high rc anyway, and
then lots of active current feedback. I don't remember the exact
configuration, but it was all discretes anyway. I'm sure you can do much
better with op-amps.

 

[Fred Bartoli]

I want a constant current generator that will produce 1 to 2 mA over an output
range of 2 to 6 volts running from a 10 to 15 volt supply.

It seems quite an easy task but here are the full requirements.

The exact tolerance of the current isn't that critical. 1mA would be fine,
1.2mA would be fine 1.6mA would be fine. Whatever is settled on, the tolerance
between different units could be as high as 10% without it causing any
problems.

Drift over time, in the long term, or with temperature isn't a problem either.
Even wild drift of up to 20% can be tolerated.

The difficult part is that over a 20mS period, and with the load varying such
that the CC output will swing from 2 to 6 volts, the current must not vary by
more than 0.05%

I'm up against a wall trying to come up with a circuit that meets this spec. No
doubt one of you lot knows one off the top of your head.

Gibbo

Chris,

If your 10/15V is stable enough during this 20ms period this will do the job
very well :


|
47K |
___ |
+15V -|___|----. |
| ||-+
| ||<- 2N7002
+---||-+
| |
' |
\| |
whatever |----+
<| |
| .-.
| | |
| | |470
| '-'
| |
| |
=== ===
GND GND
created by Andy´s ASCII-Circuit v1.22.310103 Beta www.tech-chat.de

Sensitivy from the 15V supply will be about 5uA/V though so you'll need less
than 10mV variation on your +15V.

Ha ha, source current. I bet you can do it

 

[ChrisGibboGibson]

By a couple orders of magnitude!


What do the sims show? Where is the error?

About 0.1% variation in current from 2 to 6 volts with a 10V supply. At 15V the
error is less.

Gibbo

 

[ChrisGibboGibson]

Chris,

If your 10/15V is stable enough during this 20ms period this will do the job
very well :


|
47K |
___ |
+15V -|___|----. |
| ||-+
| ||<- 2N7002
+---||-+
| |
' |
\| |
whatever |----+
<| |
| .-.
| | |
| | |470
| '-'
| |
| |
=== ===
GND GND
created by Andy´s ASCII-Circuit v1.22.310103 Beta www.tech-chat.de

Sensitivy from the 15V supply will be about 5uA/V though so you'll need less
than 10mV variation on your +15V.

Ha ha, source current. I bet you can do it

Thanks. I'll turn it upside down tomorrow and try it as a source. I can ensure
the 10/15V is within 10mV quite easily.

Looks too easy to true to me though

Gibbo

 

[Fred Bartoli]

Thanks. I'll turn it upside down tomorrow and try it as a source. I can ensure
the 10/15V is within 10mV quite easily.

Looks too easy to true to me though

Yeah, just simulated it and gives you 6nA stability for 6V supply variation.
This is an easy 1Gohm

You can reduce the 15V sensitivity by a 10-15 time factor at the expense of
an additionnal transistor :


|
10K 2.2K |
___ ___ |
+15V -|___|--+---|___|----. |
| | ||-+

| | ||<-

|----------+---||-+
/| | |
| ' |
| \| |
=== |----+
GND <| |
| .-.
| | |
| | |470
| '-'
| |
| |
=== ===
GND GND

created by Andy´s ASCII-Circuit v1.22.310103 Beta www.tech-chat.de

This has also the nice effect of increasing the output resistor by almost
three time (loop gain is increased by the constant current collector load).

 

[Fred Bartoli]

Looks too easy to true to me though

Nah, the "trick" is to use a mosfet to avoid the output transistor base
current variation due to base width modulation.

 

[ChrisGibboGibson]

Yeah, just simulated it and gives you 6nA stability for 6V supply variation.
This is an easy 1Gohm

You can reduce the 15V sensitivity by a 10-15 time factor at the expense of
an additionnal transistor :


|
10K 2.2K |
___ ___ |
+15V -|___|--+---|___|----. |
| | ||-+
|----------+---||-+
/| | |
| ' |
| \| |
=== |----+
GND <| |
| .-.
| | |
| | |470
| '-'
| |
| |
=== ===
GND GND

created by Andy´s ASCII-Circuit v1.22.310103 Beta www.tech-chat.de

This has also the nice effect of increasing the output resistor by almost
three time (loop gain is increased by the constant current collector load).

That's a neat trick but I just simulated the first one and that's plenty good
enough.

Cheers

Gibbo

 

[Jim Thompson]

On Fri, 1 Oct 2004 01:16:08 +0200, "Fred Bartoli"

[snip][snip]

Replace "whatever" NPN with a TL431 or the low-voltage (1.25V)
equivalent.

...Jim Thompson

 

[Tony Williams]

I want a constant current generator that will produce 1 to 2 mA
over an output range of 2 to 6 volts running from a 10 to 15
volt supply.

The difficult part is that over a 20mS period, and with the load
varying such that the CC output will swing from 2 to 6 volts, the
current must not vary by more than 0.05%

A CC with a common-base output is worth looking at.

15V------+----+----------+
| | |
| \_|_1N750 \
| /_\ 4v7 / 3k6
| | |
| | |/e
0.1 === +--------|pnp,2N3906
| | |\
| _|_ |
| \_/ |
| | |/e
+----+--------|pnp,2N3906
| |\
\ |
2k2 / |
\ +----> Iout, 1.09mA approx.
|
0V-----------+---------------<

A quick diddle with LTSpice, with a 2-6V 75Hz square wave
voltage generator on the Iout terminal, shows a dc-current
change of about 0.2uA. That's about 0.02%.

There are some nice spikes to get rid of though. I'd take
the easy route and slap a ferrite bead inductor in series
with Iout.

 

[Rene Tschaggelar]

John Larkin wrote:


About 0.1% variation in current from 2 to 6 volts with a 10V supply. At 15V the
error is less.

The DC error is limited by the open loop gain of the opamp.

Rene

 

[Winfield Hill]

Jim Thompson wrote...

Replace "whatever" NPN with a TL431 or the low-voltage (1.25V)
equivalent.

| +V |
| | |
| 100k ||-' 2N7000
| | ||<- 2N7002
| +---||-+
| \__|__ | 1.23
| / \_____| I = ---- = 1.0 mA
| LM385-adj /___\ | R R
| | 1.2k
| | |
| === ===
| GND GND

 

[ChrisGibboGibson]

Jim Thompson wrote...

| +V |
| | |
| 100k ||-' 2N7000
| | ||<- 2N7002
| +---||-+
| \__|__ | 1.23
| / \_____| I = ---- = 1.0 mA
| LM385-adj /___\ | R R
| | 1.2k
| | |
| === ===
| GND GND


--
Thanks,
- Win

(email: use hill_at_rowland-dotties-org for now)

I'll go work out how to turn it into a source instead of a sink.

Thanks
Gibbo

 

[Fred Bartoli]

Jim Thompson wrote...

| +V |
| | |
| 100k ||-' 2N7000
| | ||<- 2N7002
| +---||-+
| \__|__ | 1.23
| / \_____| I = ---- = 1.0 mA
| LM385-adj /___\ | R R
| | 1.2k
| | |
| === ===
| GND GND

Hey, is it a componant count contest, or transistor count contest ?

We have the same count on the first one, I win on the second

Oh... What Chris need is a current *source*. I win the third
Well, hum, unless you can use the 385/431 in reverse or have a nice trick to
share.

 

[ChrisGibboGibson]

Hey, is it a componant count contest, or transistor count contest ?

We have the same count on the first one, I win on the second

Oh... What Chris need is a current *source*. I win the third
Well, hum, unless you can use the 385/431 in reverse or have a nice trick to
share.

Yeah I'm still working on that and making no progress.

Fred wins so far.

Gibbo

 

[John Larkin]

The DC error is limited by the open loop gain of the opamp.

To, maybe, a few PPM.

John

 

[Rene Tschaggelar]

To, maybe, a few PPM.

I'd weigh the AC performance to be more interesting.
The proposed 'precision amplifiers' with DC gains of 140dB
are not that bad for lower frequency AC though.
The faster amplifiers have worse offset behaviour, but the
zero offset combo amplifiers seen in all datasheets could
perhaps improve that.
Just in case speed is an issue : a current sink with a mosfet
becomes saturated when the current is not there. Meaning the
driving amplifier goes to rail. This is not the case with
an NPN. When the current is not there, the driving amplifier
tries to push this current through the base.

Rene