Author Topic: Choosing an OPAMP for powering other opamps.  (Read 2941 times)

0 Members and 1 Guest are viewing this topic.

Offline npelov

  • Frequent Contributor
  • **
  • Posts: 290
  • Country: bg
    • Microlab.info
Choosing an OPAMP for powering other opamps.
« on: May 20, 2015, 10:17:29 am »
Hi,

I want to use 5V opamps in +/- 12V split supply power supplies. First thing I can think of to power them is 2 opamps that can take >+/-12V power supply and use them to buffer and invert a 2.5 voltage reference. My question is what opamp should I choose for maximum stability? Let's say that I will keep the current less than 10 mA. Will a jelly bean LM358 do the job just fine?
 

Offline blueskull

  • Supporter
  • ****
  • !
  • Posts: 12885
  • Country: cn
  • Power Electronics Guy
Re: Choosing an OPAMP for powering other opamps.
« Reply #1 on: May 20, 2015, 10:28:56 am »
Try this.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 14547
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Choosing an OPAMP for powering other opamps.
« Reply #2 on: May 20, 2015, 11:25:14 am »
Why would you want to use 5V amps in a 24V circuit in the first place?
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline jeroen79

  • Frequent Contributor
  • **
  • Posts: 503
Re: Choosing an OPAMP for powering other opamps.
« Reply #3 on: May 20, 2015, 11:49:16 am »
And why not use a 7805 regulator?
It'll take up less space and will require less components.
 

Offline Asmyldof

  • Regular Contributor
  • *
  • Posts: 148
  • Country: nl
  • Freelancer - Persnicketist - Do'er of stuff
    • Asmyldof's Home. It's old, not quite impressive, but it's there.
Re: Choosing an OPAMP for powering other opamps.
« Reply #4 on: May 20, 2015, 11:52:38 am »
Assuming you have considered your problem and have good reasons, such as specific performance versus cost, these things do happen in the real world.

Sometimes one sees Audio Op-Amps abused for this.
Basically what you have to consider in the decision you make is:
  • Maximum supply voltages (go at least 20% over your peak voltage, so for +13V maximum, use +15.6V or more as a requirement on the positive rail)
  • Maximum output current as a function of the output voltage drop (with going from +12V supply to +2.5V output you'll be fine on that for any amp)
  • Maximum power dissipation - Never forget you are "heating up" at 10V times the output current + some overhead, that's not quite little.

Be aware that you need to do this for your worst case scenario: When you are drawing most current, with highest voltages at the highest possible environment temperature.

Another solution, if your reference can source the required ground leakage of a regulator is using a positive and negative V-reg in cascade.
Let's say we chose a 7905 and a 7805, because they are well known types, but there are better choices, types with lower quiescent current will have a lower ground-pin current.

You hook the ground of the 7905 up to your 2.5V ref, and its input to the negative rail. The output will now make 5V less than its ground pin, which is -2.5V. If you then hook up the 7805's ground pin to the output of the 7905, which is -2.5V, so the 7805 will create 5V more than its ground pin, which is 2.5V again, but with actual oomph behind it.

Now your reference is always sourcing at least the quiescent current of the 7905 and at worst case some more (the maximum number will be noted in the regulator datasheet). So you need to be 100% sure your reference can handle that and still be accurate enough. Otherwise you could again buffer that reference with an op-amp, but at some point in that process you will be snowballing your complexity unnecessarily.
« Last Edit: May 20, 2015, 11:55:14 am by Asmyldof »
If it's a puzzle, I want to solve it.
If it's a problem, I need to solve it.
If it's an equation... mjeh, I've got Matlab
...
...
(not really though, Matlab annoys me).
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 1014
  • Country: us
  • Retired, now restoring antique test equipment
Re: Choosing an OPAMP for powering other opamps.
« Reply #5 on: May 20, 2015, 12:26:42 pm »
Do you need extremely stable power to the op amp?  Otherwise, you could use series zeners to drop the voltage, which would dissipate slightly less power than a three-terminal regulator or op amp circuit.
 

Offline npelov

  • Frequent Contributor
  • **
  • Posts: 290
  • Country: bg
    • Microlab.info
Re: Choosing an OPAMP for powering other opamps.
« Reply #6 on: May 20, 2015, 01:20:00 pm »
I'm thinking to use the ~2.4V reference of MCP3901. The problem is that it's quite high impedance - 7k or something. So in any case I'll have to buffer that with an opamp -> I can use it for positive supply. For negative - I haven't seen that many negative voltage regulators. 79L05 is drawing 3mA -> that's probably more than all opamps that will be powered. And it's easier to get a dual opamp for the buffer and use one of them as inverting amp with gain = 1 for the negative.

@Asmyldof
Yes, 10V drop is a lot. Having the opamp quiescent current - about 1mA (12*1mA = 12mW), plus, let's say 2*5mA*10V = 100mW so the dual opamp shouldn't be dissipating more than 120mW.

@jeroen79
Because I need the positive and negative rails centered around ground. So I need +2 to +2.5 positive and -2 to -2.5 negative voltage.

Why would you want to use 5V amps in a 24V circuit in the first place?
Good question! First of all I need a chopper amp for current sense. I watched Dave's videos of uCurrent and I thought I have 2 options - either get a chopper amp or do auto offset voltage compensation periodically. So I chose the easier way - chopper amp and I couldn't find one that can take higher supply voltage. Currently I'm not sure if I need the +/-12V either. In any case I'll need about +/-2 V around ground.

But if you suggest a good chopper amp that works with at least +/-9V it would be nice. The biggest problem would be offset drift with temperature and over time, because I'll use it with high gain for current sensing. Initial value of the voltage offset won't be a problem - I can null it in software.

Ow, and if I get rid of the 5V amps I'll need some very low input bias current opamps, so I can use 10MOhm voltage divider. In 5V I would use microchip's MCP601 to MCP609.
« Last Edit: May 20, 2015, 01:37:22 pm by npelov »
 

Offline flynwill

  • Regular Contributor
  • *
  • Posts: 141
Re: Choosing an OPAMP for powering other opamps.
« Reply #7 on: May 20, 2015, 01:56:30 pm »
Have you tried using the "parametric search" feature on analog device's web site:

I don't know for sure if this link will work: http://www.analog.com/parametricsearch/en/10285#10285/p299=95|50000000

But if not its not hard to get there from their top page.  You are looking for "Precision Zero Drift/Low Tempco Amplifiers"

 

Offline npelov

  • Frequent Contributor
  • **
  • Posts: 290
  • Country: bg
    • Microlab.info
Re: Choosing an OPAMP for powering other opamps.
« Reply #8 on: May 20, 2015, 02:14:51 pm »
Have you tried using the "parametric search" feature on analog device's web site:
Thanks. I wish mouser/digikey had a search like this. Most of the time they have numeric values as strings and often includes units - like 0.1uV and 100nV are different things.

... well the price in this search seams to have non-linear relation with 1 to 10 pcs prices. The cheaper opamps are two times more expensive in small quantities in mouser/digikey
« Last Edit: May 20, 2015, 03:28:03 pm by npelov »
 

Offline Asmyldof

  • Regular Contributor
  • *
  • Posts: 148
  • Country: nl
  • Freelancer - Persnicketist - Do'er of stuff
    • Asmyldof's Home. It's old, not quite impressive, but it's there.
Re: Choosing an OPAMP for powering other opamps.
« Reply #9 on: May 20, 2015, 05:07:56 pm »
The commercial pricing system of a scaled distributor depends on many factors. How often something is needed/ordered in what quantity and how much overhead that then generates is one such factor.

If a part is (or is expected to be) ordered as a single chip very often, this is insanely expensive stock for a company, because they have to cut off one, put it in a bag, seal the bag, tag the bag, inspect the bag at a second stage (especially if the first bit is automated), adding actually much more to the unit price than you have to pay extra.
So, they have a 10pc price of $1 for example, but for 5 they are $2 and for 1 they are $5, to encourage you to spend a tiny bit more on buying 10, because that will give a much better turn-over rate on the reels they are ordering.

If, instead a part is expected to be ordered with great frequency in many different quantities, the pricing will be adjusted to reflect that expectation, with for example $1/pc at 10pcs, $1.3/pc at 5 pcs and $1.7/pc at 1pc. You still pay more, because they still have to do all that work for one chip, but because the expected turn-over rate is high, the people ordering 100's are paying part of your labour bill.
Because part of their expectation is that you order 1 for a test, and then immediately decide to order 100 more if your test worked.
« Last Edit: May 20, 2015, 05:09:47 pm by Asmyldof »
If it's a puzzle, I want to solve it.
If it's a problem, I need to solve it.
If it's an equation... mjeh, I've got Matlab
...
...
(not really though, Matlab annoys me).
 

Offline npelov

  • Frequent Contributor
  • **
  • Posts: 290
  • Country: bg
    • Microlab.info
Re: Choosing an OPAMP for powering other opamps.
« Reply #10 on: May 20, 2015, 06:14:31 pm »
@Asmyldof sounds reasonable. So when the cheaper at manufacturer chip is actually more expensive at dealer to lower quantities, it's probably one of two reasons:
1. it's not commonly used because price/quality ratio is not that great
2. it's relatively new or too old/not-for-new-designs and people don't buy it that much.
3. And very little chance  the part is specific for certain usage in non-generic equipment.

My conclusion is: don't buy parts which price is drastically higher at dealers. The only chance to be good is if it's new product, but even then - the new things often have bugs.

Well I answered my question - most voltage regulators have worse regulation and greater ripple/noise than a precision voltage reference and any opamp (for low current applications). Correct me if I'm wrong.
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 14547
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Choosing an OPAMP for powering other opamps.
« Reply #11 on: May 21, 2015, 10:26:52 am »
Why would you want to use 5V amps in a 24V circuit in the first place?
Good question! First of all I need a chopper amp for current sense. I watched Dave's videos of uCurrent and I thought I have 2 options - either get a chopper amp or do auto offset voltage compensation periodically. So I chose the easier way - chopper amp and I couldn't find one that can take higher supply voltage. Currently I'm not sure if I need the +/-12V either. In any case I'll need about +/-2 V around ground.

But if you suggest a good chopper amp that works with at least +/-9V it would be nice. The biggest problem would be offset drift with temperature and over time, because I'll use it with high gain for current sensing. Initial value of the voltage offset won't be a problem - I can null it in software.

Ow, and if I get rid of the 5V amps I'll need some very low input bias current opamps, so I can use 10MOhm voltage divider. In 5V I would use microchip's MCP601 to MCP609.

Nonono, see, why do you think you need a chopper amp -- what Vos max spec do you need?  What's the tradeoff?

There's always a spec, and there's always a tradeoff.

If it's current sense, then you can use stupidly high input bias bipolar types, which can be trimmed to better than 50uV.  Chopper amps, being CMOS, also have very low input bias, which you won't need.  Of course, if you get both, that's still fine.

You can also increase the gain of the shunt, at the expense of more power.  It's pretty normal to spend a few watts there, especially if it doesn't matter (e.g., linear supply).

Examples:
Digikey Search

You can also use resistor dividers to bring Vcm down, but this reduces differential gain, making the whole thing that much more noisy and sensitive to Vos.  So that you'd have to use a chopper amp.  Self-fullfilling prophecy.  But you have to use ppm range resistors to get the CMRR anywhere near as good, which is more expensive than even a $5 op-amp, let alone a proper current sense amp device with good CMRR and stable gain.  So it gets really messy, really fast.

Far be it for me to recommend a Maxim part, but their products place highly in this search:
http://www.digikey.com/product-detail/en/MAX9943AUA%2B/MAX9943AUA%2B-ND/2062041
50uV Vos, pretty ordinary otherwise.  Fast enough for DC current sense, but you'd want something faster (give or take how much gain you need) for use in a switching supply loop.

http://www.digikey.com/product-detail/en/MAX44248ASA%2B/MAX44248ASA%2B-ND/3829285
Relatively noisy, but no 1/f noise; low Vos, wide supply range.  Also on the slow side.  Not rail to rail. Autozero is related to chopper; you'd have to look up their patents to see the difference and mechanism.

http://www.digikey.com/product-detail/en/OPA180IDGKT/296-37995-1-ND/4948940
Rail-to-rail output, low noise, apparently a true chopper.

http://www.digikey.com/product-detail/en/TLE2037CD/296-10386-5-ND/380956
Very fast (non unity gain stable), low noise, low offset, good enough to use in a switching loop.

Now that I'm salivating, I think I shall store this list for future, uh, use...

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf