1. Home
  2. Electronics Design
  3. opamp power booster

opamp power booster

I=92m using the top circuit as one half of a bipolar power supply.

formatting link

V+ comes from a variable voltage regulator. The opamp power booster wants to oscillate at 200 to 300kHz. The 1k ohm base resistor and 1nF feedback cap are to kill the oscillations. I wanted to get a bit more current out of the circuit, which is fine till I get near the power rail. Then the 1 k base resistor causes too much voltage drop and the whole thing falls out of regulation. Reducing the base resistor to 300 ohms provides enough voltage head room, but too much gain peaking at ~300kHz. I pissed away about

1/2 of Monday trying to get it to behave. In desperation I threw a low pass filter in front of the base. (Second picture.) I figured this would make things worse, (opamp driving a capacitive load), but it seems to make it work! First can someone tell me why it wants to oscillate in the first place? And then is there a better way to =91fix=92 the circuit. (As opposed to adding the base capacitor.)

Thanks,

George H.

Reply to
George Herold
Loading thread data ...

At that frequency it's very likely to be a loop stability issue rather than the pass transistor oscillating.

You've got a 10 us time constant in the feedback and probably another one from the output resistance of the pass transistor times the sum of all the bypasses, so you're marginally stable. Making either one faster or slower will improve stability.

Your lowpass filter adds another pole, but also stiffens the output by a factor of 10, so it moves the Rout*Cbypass pole away from the trouble area.

Your -15V supply could be involved as well--7900-series regulators aren't the most stable things in the world. A few hundred microfarads on the -15V rail is always comforting.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058

email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net
Reply to
Phil Hobbs

LT1013 is marginally stable at the unity gain. Increase gain or use a different opamp.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

formatting link

Reply to
Vladimir Vassilevsky

With ~60° of phase margin at 0dB of gain?

That's not the problem.

...Jim Thompson

--
| 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     |
             
         Obama is not the solution, Obama is the problem!
Reply to
Jim Thompson

F
e

age

the

.

It's a big TO-220 transistor (TIP32). The spec sheet quotes a current gain bandwidth product of 3 MHz. I've got a current gain of maybe 30, so could it be the transistor? Geesh, I never looked at that spec... I'll have to try a different transistor.

r

That's what I thought, but screwing around with capacitor values didn't change the frequency of the gain peak. With no feedback cap I had a nice oscillator. Adding C's up to 1 nF, stopped the oscillations and then reduced the gain peak, But anything greater than 1nF doesn't make it any better.

a.

Yeah, I thought it could be some sort of power supply bypassing and hung extra caps here and there. (No change)

Thanks Phil, though you dismissed it, I think it might be the transistor.

George H.

t- Hide quoted text -

Reply to
George Herold

F
e

ge

he

.

text -

Thanks Vladimir, I tried a different opamp (opa2134) and that seems to make it much better. There's still this gain peak out at ~300kHz, but I don't need such large cap values to 'beat it down'. (100pF of feed back is now enough.)

George H.

Reply to
George Herold

Wow, a 3 MHz f_T? What is it, an alloy-junction germanium? With that sort of loading, I was expecting something like a 2N4403. If you're way below I_Cmax, f_T will be significantly lower than the spec. So that's three integrators in the loop!

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058

email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net
Reply to
Phil Hobbs

-

Sounds like power supply rejection rather than a gain-bandwidth issue. The LT1013 certainly is not sterling in PSRR. ...Jim Thompson

--
| 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     |
             
         Obama is not the solution, Obama is the problem!
Reply to
Jim Thompson

r
1nF

ore

nd

voltage

ut

ed

e in the

t
,

ster

a

area.

ds

at

y

Yeah I don't have any other big pnp's to try. I'll put in a 2N4403 and see what that looks like.

Could it be the cause of the 300kHz gain peak?

George H.

t- Hide quoted text -

Reply to
George Herold

F
e

d text -

=A0 =A0 ...Jim Thompson

=A0 =A0| =A0 =A0mens =A0 =A0 |

=A0 | =A0 =A0 et =A0 =A0 =A0|

=A0|

=A0 =A0 =A0 |

quoted text -

OK I'll check that out also Jim. (Keep the LT1013, but increase the gain.)

George H.

Reply to
George Herold

text -

George, You misread me. LT1013 should be unity gain stable, with ease... its phase margin is 60°. BUT, Its poor PSRR would perhaps allow a feed back path thru the supplies.

The TIP32 is also suspect... it's a gross-sized device with large CJE and CJC.

Fire it up in LTspice and run a loop gain and phase check for the whole structure.

I presume you have used the -15V for the OpAmp Vee, what is its positive (Vcc) supply? ...Jim Thompson

--
| 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     |
             
         Obama is not the solution, Obama is the problem!
Reply to
Jim Thompson

er

1nF

more

er

out

a

but

ted text -

s

z,

f

=A0 =A0 =A0 ...Jim Thompson

=A0 =A0 =A0| =A0 =A0mens =A0 =A0 |

=A0 | =A0 =A0 et =A0 =A0 =A0|

=A0 =A0|

=A0 =A0 =A0 =A0 |

=A0|

|

ide quoted text -

Thanks Jim, I think I understood you. The test I tried was to up the gain of the opamp to 2 (by throwing away 1/2 input signal.) Then use the LT1013 in the circuit. If it's a loop stability thing then this should improve the circuit... PSRR and it should be the same.

It looks the same... Hmm I'll try adding some little 10 ohm RC's to the power lines going to the opamp.

Maybe it is a power supply thing, that would be nice.

Yup -15V and +15V

=A0 =A0 ...Jim Thompson

=A0 =A0| =A0 =A0mens =A0 =A0 |

=A0 | =A0 =A0 et =A0 =A0 =A0|

=A0|

=A0 =A0 =A0 |

quoted text -

Reply to
George Herold

text -

text -

A quick pop of your circuit into my loop gain checker shows NO loop gain/phase issues.

So...

Power supply feedback

Or...

capacitive loading on the emitter of the TIP32 ...Jim Thompson

--
| 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     |
             
         Obama is not the solution, Obama is the problem!
Reply to
Jim Thompson

e:

.

ster

nd 1nF

t more

ower

nd

about

w a

ed

, but

uotedtext -

ems

kHz,

of

e.

=A0 =A0 =A0 ...Jim Thompson

=A0 =A0 =A0| =A0 =A0mens =A0 =A0 |

=A0 =A0 | =A0 =A0 et =A0 =A0 =A0|

=A0 =A0|

=A0 =A0 =A0 =A0 |

=A0|

=A0 |

Hide quoted text -

e

=A0 =A0 ...Jim Thompson

=A0 =A0| =A0 =A0mens =A0 =A0 |

=A0 | =A0 =A0 et =A0 =A0 =A0|

=A0|

=A0 =A0 =A0 =A0 |

|

de quoted text -

That 200-300khz range is about the SRF of a power supply e-cap...

Reply to
Fred Bloggs

But, it's also the right range for the power supply regulators to be poorly regulated... so another solution is to connect the collector to unregulated power, with a helper capacitor or two (that are good for 1 MHz plus or minus a decade).

Reply to
whit3rd

issue.

Another aspect is that his transconductance is positive, so that at such large wideband current gain there is positive feedback magnetically coupled into his input stage. He needs to take a closer look at his layout with an eye towards eliminating pickup loops.

Reply to
Fred Bloggs

Power transistors have a phase shift that increases as a function of current. Try using something that may help or be a lot worse: a PNP/NPN composite darlington that acts as an NPN - more gain that may shift the phase response out of the way.

Reply to
Robert Baer

te:

y.

oster

and 1nF

it more

power

and

about

ew a

red

), but

.

quotedtext -

s

z,

f

ue.

-

he

=A0 =A0 ...Jim Thompson

=A0 =A0| =A0 =A0mens =A0 =A0 |

=A0 | =A0 =A0 et =A0 =A0 =A0|

=A0|

=A0 =A0 =A0 |

quoted text -

Thanks Jim,

I tried putting some 10 ohm resistors in the supply leads to the opamp. It did nothing to the gain peak. There was no sign of the ringing on the supply rails. On the drive home I was thinking that the reduced peak with the opa2134 may be due to the increased GBW and not the PSRR. (Dang, just looking at the spec sheet of the LT1013 again. I doesn't list a GBW spec that I can see, but it looks to be below 1MHz!)

I'll try replacing the transistor with a little 2N4403 in the morning.

George H.

Reply to
George Herold

issue.

Thanks guys, It doesn't appear to be via the power supply. The circuit gets power from a +/-15V astrodyne switching supply, through an LC filter.

George H.

Reply to
George Herold

F
e

ge

he

.

of

Yeah, I can see the frequency shift as I change the input voltage.

I hate to give up the extra voltage drop of a darlington. The current circuit with a 1k ohm base resistor only 'craps out' at the top 0.5 Volts of the range. A different pnp transitor in a TO220 would be great. Something that lots of people make.

I'm thinking a switch in opamps may be the ticket. Bury the problem with GBW.

George H.

Reply to
George Herold

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.