Current mirror with gain?

Started by Jan Panteltje September 25, 2011
For some application I need a constant current drive of a few mA,
but I only can supply some uA from the opposite rail from a high impedance DAC
output.
A single transistor would do, with a base resistor as drive,
but the gain depends on beta, and that drifts with temperature,
and from part to part.
Any ideas?





On Sun, 25 Sep 2011 18:46:12 GMT, Jan Panteltje
 wrote:

>For some application I need a constant current drive of a few mA, >but I only can supply some uA from the opposite rail from a high impedance DAC output. >A single transistor would do, with a base resistor as drive, >but the gain depends on beta, and that drifts with temperature, >and from part to part. >Any ideas? > > > >
ftp://jjlarkin.lmi.net/Ipump.JPG John
On 09/25/2011 02:46 PM, Jan Panteltje wrote:
> For some application I need a constant current drive of a few mA, > but I only can supply some uA from the opposite rail from a high impedance DAC output. > A single transistor would do, with a base resistor as drive, > but the gain depends on beta, and that drifts with temperature, > and from part to part. > Any ideas?
Well, the default approach would be an op amp driving a BJT current source--you get good high frequency performance (from the BJT) and good accuracy (from the op amp). Any special constraints? Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
On Sun, 25 Sep 2011 18:46:12 GMT, Jan Panteltje
 wrote:

>For some application I need a constant current drive of a few mA, >but I only can supply some uA from the opposite rail from a high impedance DAC output. >A single transistor would do, with a base resistor as drive, >but the gain depends on beta, and that drifts with temperature, >and from part to part. >Any ideas?
--- Like this? +V | [LOAD] | Vin>----|+\ C | >--B +--|-/ E | | +----------+ | [R] | GND -- JF
On a sunny day (Sun, 25 Sep 2011 15:06:19 -0400) it happened Phil Hobbs
 wrote in
<4E7F7BAB.4070104@electrooptical.net>:

>On 09/25/2011 02:46 PM, Jan Panteltje wrote: >> For some application I need a constant current drive of a few mA, >> but I only can supply some uA from the opposite rail from a high impedance DAC output. >> A single transistor would do, with a base resistor as drive, >> but the gain depends on beta, and that drifts with temperature, >> and from part to part. >> Any ideas? > >Well, the default approach would be an op amp driving a BJT current >source--you get good high frequency performance (from the BJT) and good >accuracy (from the op amp). Any special constraints?
Yes, maximum 2 transistors, no opamsp, only 5V supply, very liitle current.
On a sunny day (Sun, 25 Sep 2011 11:57:46 -0700) it happened John Larkin
 wrote in
:

>On Sun, 25 Sep 2011 18:46:12 GMT, Jan Panteltje > wrote: > >>For some application I need a constant current drive of a few mA, >>but I only can supply some uA from the opposite rail from a high impedance DAC output. >>A single transistor would do, with a base resistor as drive, >>but the gain depends on beta, and that drifts with temperature, >>and from part to part. >>Any ideas? >> >> >> >> > >ftp://jjlarkin.lmi.net/Ipump.JPG > >John
John, that is a very nice circuit, but I do not have the space for opamps and stuff, max 2 transistors and some R I would think.
On Sep 25, 4:14=A0pm, Jan Panteltje  wrote:
> On a sunny day (Sun, 25 Sep 2011 15:06:19 -0400) it happened Phil Hobbs > wrote in > <4E7F7BAB.4070...@electrooptical.net>: > > >On 09/25/2011 02:46 PM, Jan Panteltje wrote: > >> For some application I need a constant current drive of a few mA, > >> but I only can supply some uA from the opposite rail from a high imped=
ance DAC output.
> >> A single transistor would do, with a base resistor as drive, > >> but the gain depends on beta, and that drifts with temperature, > >> and from part to part. > >> Any ideas? > > >Well, the default approach would be an op amp driving a BJT current > >source--you get good high frequency performance (from the BJT) and good > >accuracy (from the op amp). =A0Any special constraints? > > Yes, maximum 2 transistors, no opamsp, only 5V supply, very liitle curren=
t. An SOT-23-5 op amp is the same size, runs on +5v, and it's accurate. If that's really unacceptable, you could do the "old PNP follower --> NPN stage with an emitter resistor," or the complement of that. Not as accurate, but it fits your constraints. -- Cheers, James Arthur
On Sun, 25 Sep 2011 20:17:51 GMT, Jan Panteltje
 wrote:

>On a sunny day (Sun, 25 Sep 2011 11:57:46 -0700) it happened John Larkin > wrote in >: > >>On Sun, 25 Sep 2011 18:46:12 GMT, Jan Panteltje >> wrote: >> >>>For some application I need a constant current drive of a few mA, >>>but I only can supply some uA from the opposite rail from a high impedance DAC output. >>>A single transistor would do, with a base resistor as drive, >>>but the gain depends on beta, and that drifts with temperature, >>>and from part to part. >>>Any ideas? >>> >>> >>> >>> >> >>ftp://jjlarkin.lmi.net/Ipump.JPG >> >>John > >John, that is a very nice circuit, but I do not have the space for opamps >and stuff, max 2 transistors and some R I would think.
It could be simpler if you don't need a bunch of accuracy and can afford some nonlinearity, some deadband maybe. The extreme of hackedness would involve two transistors and one resistor, and would be beta dependent on one of them. Use a beta-graded part on the high side, like BCX71 maybe. It would be an NPN current driver on the dac side and a single PNP on the high side. That would be fine if, say, there's some overall software feedback somewhere around the whole loop. John
On 9/25/2011 1:46 PM, Jan Panteltje wrote:
> For some application I need a constant current drive of a few mA, > but I only can supply some uA from the opposite rail from a high impedance DAC output. > A single transistor would do, with a base resistor as drive, > but the gain depends on beta, and that drifts with temperature, > and from part to part. > Any ideas? >
Use a classic current mirror: http://en.wikipedia.org/wiki/Current_mirror except with two PNPs in the same package such as NXP's PMBT3906YS,115 which is available from Mouser for 5 cents in a quantity of one. If you don't like those, NXP makes a lot of duals. No guarantee that the two devices are matched, but at least they are reasonably coupled thermally. Cheers, John S
Jan Panteltje a �crit :
> For some application I need a constant current drive of a few mA, > but I only can supply some uA from the opposite rail from a high impedance DAC output. > A single transistor would do, with a base resistor as drive, > but the gain depends on beta, and that drifts with temperature, > and from part to part. > Any ideas? > > > > >
If you can cope with the opamp minimum supply voltage (as low as 1V for some models), the min supply quiescent current and don't need a too high frequency response then that one is simple enough with a RR opamp: v | | | |\| Ro >----+-----|+\ ___ | | >--|___|-. | .-|-/ | | | |/| | | '---+---------' | | .-. .-. | | | | | | | | '-' '-' | | === === GND GND And be careful with the stability issues... and Ro is there have a manageable loop gain. -- Thanks, Fred.
On a sunny day (Fri, 30 Sep 2011 04:12:36 -0700 (PDT)) it happened
dagmargoodboat@yahoo.com wrote in
<60f89942-d642-4760-992e-6b0f17cebeae@g29g2000yqh.googlegroups.com>:

>On Sep 29, 3:43�pm, Jan Panteltje wrote: >> On a sunny day (Wed, 28 Sep 2011 05:44:19 -0700 (PDT)) it happened >> dagmargoodb...@yahoo.com wrote in >> <5c88b88d-4a50-46ba-aca5-9702cf210...@d18g2000yql.googlegroups.com>: >> >> >> >> >On Sep 28, 4:43 am, Jan Panteltje wrote: >> >> On a sunny day (Tue, 27 Sep 2011 02:52:29 -0700 (PDT)) it happened >> >> dagmargoodb...@yahoo.com wrote in >> >> <1b8bb301-b01e-432c-be57-ab0131849...@j19g2000yqc.googlegroups.com>: >> >> >> > +5v >> >> > -+- >> >> > | >> >> > +5v .-----. >> >> > -+- | LCD | >> >> > | '-----' >> >> > R1 |<~~~ "A" = 1.7v >> >> > | |/ >> >> > +------| Q2 Iout ~= Vin/R2 >> >> > | |>. >> >> > |<' | >> >> >Vin>-----| Q1 | >> >> > |\ R2 >> >> > | | >> >> > === === >> >> >> >--then this works fine. Just divide Vin from 0-5v to something like >> >> >0-500mV, or even 0-200mV, just like you said. Two resistors. Choos= >> >e >> >> >R2 accordingly. >> >> >> >1v across Q2 is more than plenty. >> >> >> Yes, this may be the nicest of all the circuits. >> >> >Fred's feedback makes his a lot less sensitive to r_e. �OTOH, this one >> >works to Vin ~= 0v. �Each has its advantages. >> >> >> We used to call this 'complementary emitter follower'. >> >> The temperature coefficients should cancel, >> >> but Q2 will become a bit hotter due to the higher current, >> >> >Unavoidable, in both circuits. >> >> >> maybe glue Q1 Q2 together >> >> >Sure, if it matters. �For a backlight, probably not--you can barely >> >detect/see/notice/discern a 30% brightness change, unless it changes >> >quickly. >> >> >> For my application a voltage divider would go on the input. >> >> >Yup. >> >> I tried it today wih 33 Ohm for R2 and 10k for R1, >> with 0 to + 1V input (the minimum I can get from �the DAC). >> It starts up with ever so little current, >> but it becomes non-linear near 1 V input. >> Looks like I will need the voltage divider anyways, and smaller R2. >> >> Or I could do this: >> � � � � � � � � � � � �+5 >> � � � � � � � �+5 � � �| >> � � � � � � � �| � � �LCD >> � � � � � � � Rx � � � | >> � � � � � � � �| � � � c >> 0-5 V -- 22 k ----- b � � NPN >> � � � � � � � � � � � �e BC547B >> � � � � � � � � � � � �| >> � � � � � � � � � � � /// >> That removes the deadband. >> Rx so at 0V input Ube is about 0.6V (guaranteed off). >> Makes Rx maybe 150 k. > >Or you could use a mosfet, a current-limiting resistor, and PWM the >LED for brightness control. > >That's only two parts, and it's rock-solid.
It is an idea, I am already using the {IC PWM of HV generation, and the interrupt for high priority things. But there is a timer tick at lower priority.... The timer tick is 5 ms, that would mean 5 mS on and 5 ms off for 50%, but 100 ms on and 5 ms off, for 5% on, would not be so nice I think. Or 100 ms off and 5 ms on, I am sure that would be visible flicker. I added the extra resistor, things work fine, 390 kOhm, as the DAC has a vey high output impedance. I like it, and that counts.
>-- >Cheers, >James Arthur >
On Sep 29, 3:43=A0pm, Jan Panteltje  wrote:
> On a sunny day (Wed, 28 Sep 2011 05:44:19 -0700 (PDT)) it happened > dagmargoodb...@yahoo.com wrote in > <5c88b88d-4a50-46ba-aca5-9702cf210...@d18g2000yql.googlegroups.com>: > > > > >On Sep 28, 4:43 am, Jan Panteltje wrote: > >> On a sunny day (Tue, 27 Sep 2011 02:52:29 -0700 (PDT)) it happened > >> dagmargoodb...@yahoo.com wrote in > >> <1b8bb301-b01e-432c-be57-ab0131849...@j19g2000yqc.googlegroups.com>: > > >> > +5v > >> > -+- > >> > | > >> > +5v .-----. > >> > -+- | LCD | > >> > | '-----' > >> > R1 |<~~~ "A" =3D 1.7v > >> > | |/ > >> > +------| Q2 Iout ~=3D Vin/R2 > >> > | |>. > >> > |<' | > >> >Vin>-----| Q1 | > >> > |\ R2 > >> > | | > >> > =3D=3D=3D =3D=3D=3D > > >> >--then this works fine. Just divide Vin from 0-5v to something like > >> >0-500mV, or even 0-200mV, just like you said. Two resistors. Choos=3D > >e > >> >R2 accordingly. > > >> >1v across Q2 is more than plenty. > > >> Yes, this may be the nicest of all the circuits. > > >Fred's feedback makes his a lot less sensitive to r_e. =A0OTOH, this one > >works to Vin ~=3D 0v. =A0Each has its advantages. > > >> We used to call this 'complementary emitter follower'. > >> The temperature coefficients should cancel, > >> but Q2 will become a bit hotter due to the higher current, > > >Unavoidable, in both circuits. > > >> maybe glue Q1 Q2 together > > >Sure, if it matters. =A0For a backlight, probably not--you can barely > >detect/see/notice/discern a 30% brightness change, unless it changes > >quickly. > > >> For my application a voltage divider would go on the input. > > >Yup. > > I tried it today wih 33 Ohm for R2 and 10k for R1, > with 0 to + 1V input (the minimum I can get from =A0the DAC). > It starts up with ever so little current, > but it becomes non-linear near 1 V input. > Looks like I will need the voltage divider anyways, and smaller R2. > > Or I could do this: > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0+5 > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0+5 =A0 =A0 =A0| > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =A0 =A0LCD > =A0 =A0 =A0 =A0 =A0 =A0 =A0 Rx =A0 =A0 =A0 | > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =A0 =A0 c > 0-5 V -- 22 k ----- b =A0 =A0 NPN > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0e BC547B > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 /// > That removes the deadband. > Rx so at 0V input Ube is about 0.6V (guaranteed off). > Makes Rx maybe 150 k.
Or you could use a mosfet, a current-limiting resistor, and PWM the LED for brightness control. That's only two parts, and it's rock-solid. -- Cheers, James Arthur
On Wed, 28 Sep 2011 09:43:02 GMT, Jan Panteltje
 wrote:

>On a sunny day (Tue, 27 Sep 2011 02:52:29 -0700 (PDT)) it happened >dagmargoodboat@yahoo.com wrote in ><1b8bb301-b01e-432c-be57-ab0131849da6@j19g2000yqc.googlegroups.com>: > >> >> +5v >> -+- >> | >> +5v .-----. >> -+- | LCD | >> | '-----' >> R1 |<~~~ "A" =3D 1.7v >> | |/ >> +------| Q2 Iout ~=3D Vin/R2 >> | |>. >> |<' | >>Vin>-----| Q1 | >> |\ R2 >> | | >> =3D=3D=3D =3D=3D=3D >> >>--then this works fine. Just divide Vin from 0-5v to something like >>0-500mV, or even 0-200mV, just like you said. Two resistors. Choose >>R2 accordingly. >> >>1v across Q2 is more than plenty. > > >Yes, this may be the nicest of all the circuits. >We used to call this 'complementary emitter follower'. >The temperature coefficients should cancel, >but Q2 will become a bit hotter due to the higher current, >maybe glue Q1 Q2 together >For my application a voltage divider would go on the input. >
It is somewhat similar to a Szlikai pair.
On a sunny day (Wed, 28 Sep 2011 05:44:19 -0700 (PDT)) it happened
dagmargoodboat@yahoo.com wrote in
<5c88b88d-4a50-46ba-aca5-9702cf210828@d18g2000yql.googlegroups.com>:

>On Sep 28, 4:43�am, Jan Panteltje wrote: >> On a sunny day (Tue, 27 Sep 2011 02:52:29 -0700 (PDT)) it happened >> dagmargoodb...@yahoo.com wrote in >> <1b8bb301-b01e-432c-be57-ab0131849...@j19g2000yqc.googlegroups.com>: >> >> >> >> >> >> > � � � � � � � � � +5v >> > � � � � � � � � � -+- >> > � � � � � � � � � �| >> > � � � � �+5v � �.-----. >> > � � � � �-+- � �| LCD | >> > � � � � � | � � '-----' >> > � � � � �R1 � � � �|<~~~ "A" = 1.7v >> > � � � � � | � � �|/ >> > � � � � � +------| �Q2 � Iout ~= Vin/R2 >> > � � � � � | � � �|>. >> > � � � � |<' � � � �| >> >Vin>-----| �Q1 � � �| >> > � � � � |\ � � � �R2 >> > � � � � � | � � � �| >> > � � � � �=== � � �=== >> >> >--then this works fine. �Just divide Vin from 0-5v to something like >> >0-500mV, or even 0-200mV, just like you said. �Two resistors. �Choos= >e >> >R2 accordingly. >> >> >1v across Q2 is more than plenty. >> >> Yes, this may be the nicest of all the circuits. > >Fred's feedback makes his a lot less sensitive to r_e. OTOH, this one >works to Vin ~= 0v. Each has its advantages. > >> We used to call this 'complementary emitter follower'. >> The temperature coefficients should cancel, >> but Q2 will become a bit hotter due to the higher current, > >Unavoidable, in both circuits. > >> maybe glue Q1 Q2 together > >Sure, if it matters. For a backlight, probably not--you can barely >detect/see/notice/discern a 30% brightness change, unless it changes >quickly. > >> For my application a voltage divider would go on the input. > >Yup.
I tried it today wih 33 Ohm for R2 and 10k for R1, with 0 to + 1V input (the minimum I can get from the DAC). It starts up with ever so little current, but it becomes non-linear near 1 V input. Looks like I will need the voltage divider anyways, and smaller R2. Or I could do this: +5 +5 | | LCD Rx | | c 0-5 V -- 22 k ----- b NPN e BC547B | /// That removes the deadband. Rx so at 0V input Ube is about 0.6V (guaranteed off). Makes Rx maybe 150 k.
On Sep 28, 10:20=A0am, Fred Bloggs 
wrote:
> On Sep 28, 9:07=A0am, dagmargoodb...@yahoo.com wrote: > > > > > On Sep 28, 3:00=A0am, Jon Kirwan wrote: > > > > rOn Tue, 27 Sep 2011 07:35:56 -0700 (PDT), > > > > dagmargoodb...@yahoo.com wrote: > > > >On Sep 27, 8:02 am, Jon Kirwan wrote: > > > >> On Tue, 27 Sep 2011 09:00:07 GMT, Jan Panteltje > > > > >> wrote: > > > >> >On a sunny day (Mon, 26 Sep 2011 22:20:11 -0700 (PDT)) it happene=
d Fred Bloggs
> > > >> > wrote in > > > >> >m>:
> > > > >> >>Voltage drop across emitter resistors is way too small compared =
to
> > > >> >>VBEQ. > > > >> >>There is such a thing as a compound darlington, something like b=
elow-
> > > >> >>I winged the parameters, it is temperature stable and should get=
you
> > > >> >>within 2% even on a bad day- it's not a mirror: > > > >> >> Please view in a fixed-width font such as Courier. > > > . > > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0+5V
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0.-----=
-----.
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0 =A0|
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0 =A0|
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0[10k] =A0 =
=A0 =A0 =A0|
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0|< =A0 Q2
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0+-----=
-- | =A0BC557
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0|\
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0 =A0|
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
.---------------.
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
| 3,3V drop =A0 =A0 |
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
| LCD backlight |
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
'---------------'
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0 =A0|
> > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0|/ =A0 =A0=
=A0 =A0 =A0 |
> > . =A0 =A0 =A0 0-+5V =A0 >-[47k]-+------| =A0 Q1 =A0 =A0 =A0 | > > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 | =A0 =A0 =A0|> BC547B =
=A0 =A0|
> > . =A0 =A0 =A0Pin 19 =A0 =A0 =A0 =A0 [5.1K] =A0 =A0 | =A0 =A0 =A0 =A0 =
=A0|
> > . =A0 =A0 =A0PIC DAC out =A0 =A0 =A0| =A0 =A0 =A0 =A0+----------' > > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 - =A0 =A0 =A0 =A0| > > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 v =A0 =A0 =A0[15] > > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A01N914A - =A0 =A0 =A0 =A0| > > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 | =A0 =A0 =A0 =A0| > > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0/// =A0 =A0 =A0/// > > > > >> I found the bootstrapping idea for the standard degenerative > > > >> BJT amplifier, the one used to greatly stiffen the input > > > >> impedance, something quite enjoyable to learn about as well. > > > >> But it wasn't in the same league of 'simple and beautiful > > > >> elegance' as the above pattern is, to me. > > > > >One caveat--it can oscillate. =A0Happened to me for an A/D buffer, b=
ack
> > > >in the horse-and-buggy days. > > > > Although I haven't used the above form exactly (and there are > > > enough differences that it may matter) I'd appreciate it if > > > you might explain what can cause that oscillation. =A0It would > > > help me to understand it better. > > > > Jon > > > Where you have gain, feedback, and phase delay, there be dragons. > > > Let's look at it in the time domain. =A0Suppose Q1 is slow and Q2 fast. > > Imagine a small impulse into Q1(b). =A0Imagine Q1's collector current > > rising slowly, and i(Q2c)'s rise following very closely behind, > > applying negative feedback in lock-step to Q1's emitter. =A0Stable. > > > Suppose the same impulse where Q1 is fast and Q2 slow. =A0Time-wise, > > Q2's feedback signal to Q1 can't keep up. =A0i(Q1c), lacking feedback, > > overshoots. =A0Q2 finally responds. =A0Q2,wound up, then applies too mu=
ch
> > feedback. =A0And, continues to increase it even after Q1 is begging it > > not to. =A0The output overshoots. =A0Then, with Q1 overly starved (beca=
use
> > the output's swung much too high), the process reverses and > > undershoots. =A0Repeat. =A0Unstable. > > > Mine screamed at 10's or 100's of MHz (I don't remember clearly). > > > That's one example. =A0There are probably multiple possibilities. > > > -- > > Cheers, > > James Arthur >
> > Ummmm...are you serious? The circuit is dominated by only two poles , > one of which is at really high frequency because of the scant 15R > load. Loop phase shift is asymptotic to but never reaches 180o unless > you have something else going on there.
Sorry, I lazily re-published your schematic for the topology, but wasn't talking about your implementation. I should've stripped out the component values. I don't remember the combination that made my buffer scream way back when, only that it did. -- Cheers, James Arthur
On a sunny day (Wed, 28 Sep 2011 05:44:19 -0700 (PDT)) it happened
dagmargoodboat@yahoo.com wrote in
<5c88b88d-4a50-46ba-aca5-9702cf210828@d18g2000yql.googlegroups.com>:

>On Sep 28, 4:43�am, Jan Panteltje wrote: >> On a sunny day (Tue, 27 Sep 2011 02:52:29 -0700 (PDT)) it happened >> dagmargoodb...@yahoo.com wrote in >> <1b8bb301-b01e-432c-be57-ab0131849...@j19g2000yqc.googlegroups.com>: >> >> >> >> >> >> > � � � � � � � � � +5v >> > � � � � � � � � � -+- >> > � � � � � � � � � �| >> > � � � � �+5v � �.-----. >> > � � � � �-+- � �| LCD | >> > � � � � � | � � '-----' >> > � � � � �R1 � � � �|<~~~ "A" = 1.7v >> > � � � � � | � � �|/ >> > � � � � � +------| �Q2 � Iout ~= Vin/R2 >> > � � � � � | � � �|>. >> > � � � � |<' � � � �| >> >Vin>-----| �Q1 � � �| >> > � � � � |\ � � � �R2 >> > � � � � � | � � � �| >> > � � � � �=== � � �=== >> >> >--then this works fine. �Just divide Vin from 0-5v to something like >> >0-500mV, or even 0-200mV, just like you said. �Two resistors. �Choos= >e >> >R2 accordingly. >> >> >1v across Q2 is more than plenty. >> >> Yes, this may be the nicest of all the circuits. > >Fred's feedback makes his a lot less sensitive to r_e. OTOH, this one >works to Vin ~= 0v. Each has its advantages. > >> We used to call this 'complementary emitter follower'. >> The temperature coefficients should cancel, >> but Q2 will become a bit hotter due to the higher current, > >Unavoidable, in both circuits. > >> maybe glue Q1 Q2 together > >Sure, if it matters. For a backlight, probably not--you can barely >detect/see/notice/discern a 30% brightness change, unless it changes >quickly. > >> For my application a voltage divider would go on the input. > >Yup.
Sure, that backlight is not critical. I tell you what, I realised today I can omit the voltage divider, as the PIC has a choice to set the reference of the DAC to its internal reference (stabilised too) of 1.024 V. This gives me 32 steps of 32 mV. So for 30 mA maximum that makes R2 = 1.024 / .03 ~ 33 Ohm. It is also nicely below the max voltage drop allowed, so plenty of voltage for the LCD backlight. I will try this later, now working on an other part of the project, This circuit now requires only 2 transistors and 2 resistors :-)
On Sep 28, 9:07=A0am, dagmargoodb...@yahoo.com wrote:
> On Sep 28, 3:00=A0am, Jon Kirwan wrote: > > > > > rOn Tue, 27 Sep 2011 07:35:56 -0700 (PDT), > > > dagmargoodb...@yahoo.com wrote: > > >On Sep 27, 8:02 am, Jon Kirwan wrote: > > >> On Tue, 27 Sep 2011 09:00:07 GMT, Jan Panteltje > > > >> wrote: > > >> >On a sunny day (Mon, 26 Sep 2011 22:20:11 -0700 (PDT)) it happened =
Fred Bloggs
> > >> > wrote in > > >> >=
:
> > > >> >>Voltage drop across emitter resistors is way too small compared to > > >> >>VBEQ. > > >> >>There is such a thing as a compound darlington, something like bel=
ow-
> > >> >>I winged the parameters, it is temperature stable and should get y=
ou
> > >> >>within 2% even on a bad day- it's not a mirror: > > >> >> Please view in a fixed-width font such as Courier. > > . > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0=
+5V
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0.-------=
---.
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0 =A0|
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0 =A0|
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0[10k] =A0 =
=A0 =A0 =A0|
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0|< =A0 Q2
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0+-------=
| =A0BC557
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0|\
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0 =A0|
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 .-=
--------------.
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 | =
3,3V drop =A0 =A0 |
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 | =
LCD backlight |
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 '-=
--------------'
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| =A0 =
=A0 =A0 =A0 =A0|
> . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0|/ =A0 =A0 =
=A0 =A0 =A0 |
> . =A0 =A0 =A0 0-+5V =A0 >-[47k]-+------| =A0 Q1 =A0 =A0 =A0 | > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 | =A0 =A0 =A0|> BC547B =A0 =
=A0|
> . =A0 =A0 =A0Pin 19 =A0 =A0 =A0 =A0 [5.1K] =A0 =A0 | =A0 =A0 =A0 =A0 =A0| > . =A0 =A0 =A0PIC DAC out =A0 =A0 =A0| =A0 =A0 =A0 =A0+----------' > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 - =A0 =A0 =A0 =A0| > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 v =A0 =A0 =A0[15] > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A01N914A - =A0 =A0 =A0 =A0| > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 | =A0 =A0 =A0 =A0| > . =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0/// =A0 =A0 =A0/// > > > > > > > > > >> I found the bootstrapping idea for the standard degenerative > > >> BJT amplifier, the one used to greatly stiffen the input > > >> impedance, something quite enjoyable to learn about as well. > > >> But it wasn't in the same league of 'simple and beautiful > > >> elegance' as the above pattern is, to me. > > > >One caveat--it can oscillate. =A0Happened to me for an A/D buffer, bac=
k
> > >in the horse-and-buggy days. > > > Although I haven't used the above form exactly (and there are > > enough differences that it may matter) I'd appreciate it if > > you might explain what can cause that oscillation. =A0It would > > help me to understand it better. > > > Jon > > Where you have gain, feedback, and phase delay, there be dragons. > > Let's look at it in the time domain. =A0Suppose Q1 is slow and Q2 fast. > Imagine a small impulse into Q1(b). =A0Imagine Q1's collector current > rising slowly, and i(Q2c)'s rise following very closely behind, > applying negative feedback in lock-step to Q1's emitter. =A0Stable. > > Suppose the same impulse where Q1 is fast and Q2 slow. =A0Time-wise, > Q2's feedback signal to Q1 can't keep up. =A0i(Q1c), lacking feedback, > overshoots. =A0Q2 finally responds. =A0Q2,wound up, then applies too much > feedback. =A0And, continues to increase it even after Q1 is begging it > not to. =A0The output overshoots. =A0Then, with Q1 overly starved (becaus=
e
> the output's swung much too high), the process reverses and > undershoots. =A0Repeat. =A0Unstable. > > Mine screamed at 10's or 100's of MHz (I don't remember clearly). > > That's one example. =A0There are probably multiple possibilities. > > -- > Cheers, > James Arthur- Hide quoted text - > > - Show quoted text -- Hide quoted text - > > - Show quoted text -
Ummmm...are you serious? The circuit is dominated by only two poles , one of which is at really high frequency because of the scant 15R load. Loop phase shift is asymptotic to but never reaches 180o unless you have something else going on there.
On Sep 28, 3:00=A0am, Jon Kirwan  wrote:
> rOn Tue, 27 Sep 2011 07:35:56 -0700 (PDT), > > > > dagmargoodb...@yahoo.com wrote: > >On Sep 27, 8:02 am, Jon Kirwan wrote: > >> On Tue, 27 Sep 2011 09:00:07 GMT, Jan Panteltje > > >> wrote: > >> >On a sunny day (Mon, 26 Sep 2011 22:20:11 -0700 (PDT)) it happened Fr=
ed Bloggs
> >> > wrote in > >> >: > > >> >>Voltage drop across emitter resistors is way too small compared to > >> >>VBEQ. > >> >>There is such a thing as a compound darlington, something like below=
-
> >> >>I winged the parameters, it is temperature stable and should get you > >> >>within 2% even on a bad day- it's not a mirror: > >> >> Please view in a fixed-width font such as Courier.
. . +5V . .----------. . | | . | | . [10k] | . | |< Q2 . +------- | BC557 . | |\ . | | . | .---------------. . | | 3,3V drop | . | | LCD backlight | . | '---------------' . | | . |/ | . 0-+5V >-[47k]-+------| Q1 | . | |> BC547B | . Pin 19 [5.1K] | | . PIC DAC out | +----------' . - | . v [15] . 1N914A - | . | | . /// ///
> > >> I found the bootstrapping idea for the standard degenerative > >> BJT amplifier, the one used to greatly stiffen the input > >> impedance, something quite enjoyable to learn about as well. > >> But it wasn't in the same league of 'simple and beautiful > >> elegance' as the above pattern is, to me. > > >One caveat--it can oscillate. =A0Happened to me for an A/D buffer, back > >in the horse-and-buggy days. > > Although I haven't used the above form exactly (and there are > enough differences that it may matter) I'd appreciate it if > you might explain what can cause that oscillation. =A0It would > help me to understand it better. > > Jon
Where you have gain, feedback, and phase delay, there be dragons. Let's look at it in the time domain. Suppose Q1 is slow and Q2 fast. Imagine a small impulse into Q1(b). Imagine Q1's collector current rising slowly, and i(Q2c)'s rise following very closely behind, applying negative feedback in lock-step to Q1's emitter. Stable. Suppose the same impulse where Q1 is fast and Q2 slow. Time-wise, Q2's feedback signal to Q1 can't keep up. i(Q1c), lacking feedback, overshoots. Q2 finally responds. Q2,wound up, then applies too much feedback. And, continues to increase it even after Q1 is begging it not to. The output overshoots. Then, with Q1 overly starved (because the output's swung much too high), the process reverses and undershoots. Repeat. Unstable. Mine screamed at 10's or 100's of MHz (I don't remember clearly). That's one example. There are probably multiple possibilities. -- Cheers, James Arthur
On Sep 28, 4:43=A0am, Jan Panteltje  wrote:
> On a sunny day (Tue, 27 Sep 2011 02:52:29 -0700 (PDT)) it happened > dagmargoodb...@yahoo.com wrote in > <1b8bb301-b01e-432c-be57-ab0131849...@j19g2000yqc.googlegroups.com>: > > > > > > > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 +5v > > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 -+- > > =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0| > > =A0 =A0 =A0 =A0 =A0+5v =A0 =A0.-----. > > =A0 =A0 =A0 =A0 =A0-+- =A0 =A0| LCD | > > =A0 =A0 =A0 =A0 =A0 | =A0 =A0 '-----' > > =A0 =A0 =A0 =A0 =A0R1 =A0 =A0 =A0 =A0|<~~~ "A" =3D 1.7v > > =A0 =A0 =A0 =A0 =A0 | =A0 =A0 =A0|/ > > =A0 =A0 =A0 =A0 =A0 +------| =A0Q2 =A0 Iout ~=3D Vin/R2 > > =A0 =A0 =A0 =A0 =A0 | =A0 =A0 =A0|>. > > =A0 =A0 =A0 =A0 |<' =A0 =A0 =A0 =A0| > >Vin>-----| =A0Q1 =A0 =A0 =A0| > > =A0 =A0 =A0 =A0 |\ =A0 =A0 =A0 =A0R2 > > =A0 =A0 =A0 =A0 =A0 | =A0 =A0 =A0 =A0| > > =A0 =A0 =A0 =A0 =A0=3D=3D=3D =A0 =A0 =A0=3D=3D=3D > > >--then this works fine. =A0Just divide Vin from 0-5v to something like > >0-500mV, or even 0-200mV, just like you said. =A0Two resistors. =A0Choos=
e
> >R2 accordingly. > > >1v across Q2 is more than plenty. > > Yes, this may be the nicest of all the circuits.
Fred's feedback makes his a lot less sensitive to r_e. OTOH, this one works to Vin ~=3D 0v. Each has its advantages.
> We used to call this 'complementary emitter follower'. > The temperature coefficients should cancel, > but Q2 will become a bit hotter due to the higher current,
Unavoidable, in both circuits.
> maybe glue Q1 Q2 together
Sure, if it matters. For a backlight, probably not--you can barely detect/see/notice/discern a 30% brightness change, unless it changes quickly.
> For my application a voltage divider would go on the input.
Yup. -- Cheers, James Arthur
On Wed, 28 Sep 2011 01:00:00 -0700, Jon Kirwan 
wrote:

>rOn Tue, 27 Sep 2011 07:35:56 -0700 (PDT), >dagmargoodboat@yahoo.com wrote: > >>On Sep 27, 8:02�am, Jon Kirwan wrote: >>> On Tue, 27 Sep 2011 09:00:07 GMT, Jan Panteltje >>> >>> >>> >>> wrote: >>> >On a sunny day (Mon, 26 Sep 2011 22:20:11 -0700 (PDT)) it happened Fred Bloggs >>> > wrote in >>> >: >>> >>> >>Voltage drop across emitter resistors is way too small compared to >>> >>VBEQ. >>> >>There is such a thing as a compound darlington, something like below- >>> >>I winged the parameters, it is temperature stable and should get you >>> >>within 2% even on a bad day- it's not a mirror: >>> >> �Please view in a fixed-width font such as Courier. >>> >>> >>. >>> >>. >>> >>. � � � � � � � � � � � � � � � � � �+5V >>> >>. � � � � � � � � � � � � � � � �.----------. >>> >>. � � � � � � � � � � � � � � � �| � � � � �| >>> >>. � � � � � � � � � � � � � � � �| � � � � �| >>> >>. � � � � � � � � � � � � � � �[10k] � � � �| >>> >>. � � � � � � � � � � � � � � � �| � � � �|< >>> >>. � � � � � � � � � � � � � � � �+------- | �BC557 >>> >>. � � � � � � � � � � � � � � � �| � � � �|\ >>> >>. � � � � � � � � � � � � � � � �| � � � � �| >>> >>. � � � � � � � � � � � � � � � �| � .---------------. >>> >>. � � � � � � � � � � � � � � � �| � | 3,3V drop � � | >>> >>. � � � � � � � � � � � � � � � �| � | LCD backlight | >>> >>. � � � � � � � � � � � � � � � �| � '---------------' >>> >>. � � � � � � � � � � � � � � � �| � � � � �| >>> >>. � � � � � � � � � � � � � � �|/ � � � � � | >>> >>. � � � 0-+5V � >-[47k]-+------| � � � � � �| >>> >>. � � � � � � � � � � � | � � �|> BC547B � �| >>> >>. � � �Pin 19 � � � � [5.1K] � � | � � � � �| >>> >>. � � �PIC DAC out � � �| � � � �+----------' >>> >>. � � � � � � � � � � � - � � � �| >>> >>. � � � � � � � � � � � v � � �[15] >>> >>. � � � � � � � �1N914A - � � � �| >>> >>. � � � � � � � � � � � | � � � �| >>> >>. � � � � � � � � � � �/// � � �/// >>> >>> >That is a very nice circuit, and I actually understand it (unlike the previous one) :-) >>> >Yes, this is nice, I will test this out. many thanks :-) >>> >>> I've used it. �It is excellent, not at all difficult to >>> design fairly well before testing, works remarkably well, and >>> can be readily combined with other concepts. >>> >>> Nicely, if you socket the backlight and someone accidentally >>> touches a screwdriver to the LED pins and shorts them out, it >>> doesn't behave badly as other circuits often do. �It's very >>> useful and understandbale besides. �That emitter resistor >>> use/re-use actually qualifies as downright beautiful. >>> >>> I found the bootstrapping idea for the standard degenerative >>> BJT amplifier, the one used to greatly stiffen the input >>> impedance, something quite enjoyable to learn about as well. >>> But it wasn't in the same league of 'simple and beautiful >>> elegance' as the above pattern is, to me. >> >>One caveat--it can oscillate. Happened to me for an A/D buffer, back >>in the horse-and-buggy days. > >Although I haven't used the above form exactly (and there are >enough differences that it may matter) I'd appreciate it if >you might explain what can cause that oscillation. It would >help me to understand it better. > >Jon
It IS a feedback loop. More prone to oscillate if BJT, but CMOS can be coerced :-) ...Jim Thompson -- [On the Road, in New York] | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | I love to cook with wine. Sometimes I even put it in the food.