Current Boost for Op Amps

[Rick C]

It is hard to find an op amp with good drive and a reasonable GBW product without a high idle current.

I was looking at an old design that might get a do-over because of an obsolete part and I wanted to consider replacing the op amp because of the cost. I didn't find anything at Digikey or Mouser that fits the bill any better than the current part, an LM8272. The problem is the need for small size and high current output.

This part will drive the output to Âą7.2 volts into a 100 ohm differential load (~72 mAp) before clipping using a 12 volt supply. This is in a circuit with synthetic output impedance using a 12.1 ohm output resistor. The op amp output is 9 Vpp before the output resistor. So a replacement has to give at least this much range.

The circuit is AC coupled, so the offset is not so important. The circuit needs to work up to 20 kHz so a GBW of a couple MHz would be good. I'd like to keep the supply current below 3 mA per channel and the cost below $2@1000.

Board space is extremely tight. I used dual parts in MSOPs and something at least that dense would be needed. A quad TSSOP or QFN might be good. I just haven't found anything that will drive the current without driving up the cost or the supply current.

--

Rick C.

- Get 10,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209

 

[sea moss]

I've been using the ADA4661 recently and it fits what you're going for. Probably a little "nicer / more expensive" than what you need... $1.50 in quantities of 1500 at Digikey. Dual, 18V, 3mm QFN with thermal pad.

That's assuming you don't have room for the traditional solution for current boost: add an NPN-PNP push/pull output, and take the op amp feedback from the output of the push/pull.

 

[Rick C]

On Wednesday, March 18, 2020 at 10:37:22 PM UTC-4, sea moss wrote:

I've been using the ADA4661 recently and it fits what you're going for. Probably a little "nicer / more expensive" than what you need... $1.50 in quantities of 1500 at Digikey. Dual, 18V, 3mm QFN with thermal pad.

That's assuming you don't have room for the traditional solution for current boost: add an NPN-PNP push/pull output, and take the op amp feedback from the output of the push/pull.

It looks like a great part, but for some reason when I plugged it into my simulation it was clipping badly at a point the LM8272 was just beginning to clip. There is one number in the data sheet I don't get.

Continuous Output Current IOUT Dropout voltage = 1 V 40mA

This is the same rating for 10 and 18 volts Vsy so clearly 12 volt operation would be the same. The typical graphs show the output reaching within 2 volts of the high rail at 90mA, but in the simulation the output clips hard at that voltage at less than 66 mA.

It also doesn't handle capacitive loads so I'd have to deal with that. This circuit drives cables. The LM8272 is rated to be stable for all capacitive loads.

At least they had simulation models in LTspice so it was easy to check.

Thanks for the suggestion. It does look like a nice part.

--

Rick C.

+ Get 2,000 miles of free Supercharging
+ Tesla referral code - https://ts.la/richard11209

 

[Rick C]

On Wednesday, March 18, 2020 at 10:37:22 PM UTC-4, sea moss wrote:

I've been using the ADA4661 recently and it fits what you're going for. Probably a little "nicer / more expensive" than what you need... $1.50 in quantities of 1500 at Digikey. Dual, 18V, 3mm QFN with thermal pad.

That's assuming you don't have room for the traditional solution for current boost: add an NPN-PNP push/pull output, and take the op amp feedback from the output of the push/pull.

Oh yeah, you are right that I don't have space on the board for output transistors. I also don't know how easy it would be to work those into my circuit. I guess they would just be on the op amp output before any of the feedback and the output resistor. But this board is so tight, it would be a real pain (in the literal sense of the word) to try to squeeze in 8 more parts, even tiny ones. This is already 5 kg of circuit in a 2.5 kg board.

--

Rick C.

-- Get 2,000 miles of free Supercharging
-- Tesla referral code - https://ts.la/richard11209

 

[Winfield Hill]

Rick C wrote...

Oh yeah, you are right that I don't have space on the board
for output transistors. ... I guess they would just be on
the op amp output before any of the feedback and the output
resistor. ...

I use a resistor from the op-amp output to the load, sized
for 20 to 50mA at 0.6 volts, and add npn and pnp emitter
followers around the resistor. Small sot-89 parts can do
the job. E.g., dnls350y and dpls350y. You can put these
three parts on the bottom.


--
Thanks,
- Win

 

On 19 Mar 2020 02:47:50 -0700, Winfield Hill <winfieldhill@yahoo.com>
wrote:

Rick C wrote...

Oh yeah, you are right that I don't have space on the board
for output transistors. ... I guess they would just be on
the op amp output before any of the feedback and the output
resistor. ...

I use a resistor from the op-amp output to the load, sized
for 20 to 50mA at 0.6 volts, and add npn and pnp emitter
followers around the resistor. Small sot-89 parts can do
the job. E.g., dnls350y and dpls350y. You can put these
three parts on the bottom.

Pure class-B, without the crossover-killer resistor, is
counter-intuitive and uncomfortable but seems to usually work.




--

John Larkin Highland Technology, Inc

The cork popped merrily, and Lord Peter rose to his feet.
"Bunter", he said, "I give you a toast. The triumph of Instinct over Reason"

 

[Rick C]

On Thursday, March 19, 2020 at 5:48:03 AM UTC-4, Winfield Hill wrote:

Rick C wrote...

Oh yeah, you are right that I don't have space on the board
for output transistors. ... I guess they would just be on
the op amp output before any of the feedback and the output
resistor. ...

I use a resistor from the op-amp output to the load, sized
for 20 to 50mA at 0.6 volts, and add npn and pnp emitter
followers around the resistor. Small sot-89 parts can do
the job. E.g., dnls350y and dpls350y. You can put these
three parts on the bottom.

I've already said I don't have space on the board. I'm not clear on your description "around the resistor". I may not be able to use the circuit on this board, but I'd like to understand it.

--

Rick C.

-+ Get 1,000 miles of free Supercharging
-+ Tesla referral code - https://ts.la/richard11209

 

[Winfield Hill]

Rick C wrote...

I'm not clear on your description "around the resistor".

In addition to the standard npn + pnp emitter followers,
add a resistor bypassing them, straight to the output.
This provides current for light loads. Select it for
600mV drop at nicely below the opamp's maximum current.

BTW, checkout small six-pin packages with one each
npn and pnp transistor. And the resistor is optional.

Are you using the very smallest packages available for
each part type?


--
Thanks,
- Win

 

[sea moss]

With tiny packages, the op amp + NPN/PNP could work.

e.g. NPN/PNP in a 2mm DFN with thermal pad: PBSS4112PANP,115

1.6mm DFN op amp: NCS2004AMUTAG

 

[Rick C]

On Thursday, March 19, 2020 at 1:28:39 PM UTC-4, sea moss wrote:

With tiny packages, the op amp + NPN/PNP could work.

e.g. NPN/PNP in a 2mm DFN with thermal pad: PBSS4112PANP,115

1.6mm DFN op amp: NCS2004AMUTAG

Thanks for the suggestiongs, but DFNs are not magic in fact, they can be less space efficient than leaded parts like MSOPs. An IC doesn't just take up space from other parts, it takes space away from routing and vias, those oh so essential vias. A part like a QFN or DFN typically are a solid block of obstruction in PCB layout. Nothing can go where the part is, no routing can go under the part and no vias can go under the part. An MSOP allows routes and vias to be run under the package with only the pads actually taking up board space. If you have a similar sized part it can go opposite and the two parts share the same space on both sides of the board with no further routing restrictions.

The MSOP pads use about 4 sq mm or 8 sq mm if you add in a 10 mil clearance.. The DFN approach would use more like 20 sq mm as the MSOP is a dual and the DFNs are singles.

The current circuit is working well. We've sold something like 10,000 units over the last decade or so. I just thought there might be a better op amp for this application by now.

The slightly extreme specs come from the IRIG-B120 application. Seems there is no spec for the electrical interface and there are systems that use up to 10 Vpp and there are 50 ohm systems. Not an easy range to drive, especially if you want both at the same time. lol I didn't make 10 Vpp, but I did manage to get 7 Vpp driving a 50 ohm load presenting a 50 ohm driving impedance with low distortion (14 Vpp differential into 100 ohms). Two channels of this on a board about 4.5 inches by 0.85 inches along with an RS-422 digital interface.

I wouldn't mind getting the output voltage range into 50 ohms higher, but that isn't a big deal at this point. I don't think they sell many for IRIG these days. Mostly they are sold for 600 ohm audio interfaces.

--

Rick C.

+- Get 2,000 miles of free Supercharging
+- Tesla referral code - https://ts.la/richard11209

 

[Winfield Hill]

Rick C wrote...

sea moss wrote:
NPN/PNP in a 2mm DFN with thermal pad: PBSS4112PANP

1.6mm DFN op amp: NCS2004AMUTAG

That opamp (and many others) is available in SC-70.
Likewise many npn+pnp combos, although you may need
the heat-removal pad.


--
Thanks,
- Win

 

[Rick C]

On Thursday, March 19, 2020 at 3:09:43 PM UTC-4, Winfield Hill wrote:

Rick C wrote...

sea moss wrote:
NPN/PNP in a 2mm DFN with thermal pad: PBSS4112PANP

1.6mm DFN op amp: NCS2004AMUTAG

That opamp (and many others) is available in SC-70.
Likewise many npn+pnp combos, although you may need
the heat-removal pad.

And the op amp with the transistor is still half again the size of the existing design even after accounting for the space between the rows of pins. The SC-70 is small enough that while I can run two traces between the rows of pads, I can't put a via there. Pads require a 10 mil clearance to anything a solder bridge might form to.

There is a reason why I used dual op amp packages. Eliminating two pins per package saves space and the MSOP package is routing friendly. I see the same issues between QFP and QFN packages. The QFP seems bigger, but all that space underneath can be used for vias and routes. The QFP is a solid block with the thermal pad. So even though it is "smaller", it is no easier to use on a board.

Most likely I won't do anything with this design. The customer is asking about some changes in the input circuit that will just be resistor value changes. I was just looking to see if there was an easy way to get more voltage swing into the minimum load. Looks not not so much.

To be fair, I checked my art work and it looks like I shaved the length of my MSOP8 pads a small amount. But I would not be comfortable doing that with an SC-70.

Does the crossover distortion show up in distortion tests? This unit it trying to stay close to "CD quality" as much as possible.

--

Rick C.

++ Get 2,000 miles of free Supercharging
++ Tesla referral code - https://ts.la/richard11209