diyAudio

diyAudio (https://www.diyaudio.com/forums/index.php)
-   Power Supplies (https://www.diyaudio.com/forums/power-supplies/)
-   -   Using OpAmps as Voltage Regulators (https://www.diyaudio.com/forums/power-supplies/351471-using-opamps-voltage-regulators.html)

Tfive 21st March 2020 09:41 PM

Using OpAmps as Voltage Regulators
 
1 Attachment(s)
Hey folks, we're currently in the process of designing a multi channel DAC centered around the ES9038Q2M in mono mode. The DAC will feature a motherboard with PCI express connectors, i2s-, i2c- and differential master-clock-distribution. The DAC boards themselves will be all mono channels with opto isolated i2s and i2c lines, diff-to-single-ended clock receiver, I/V and relay based volume control with 64 1dB steps. But i digress, that will be a topic for it's own thread once all the schematics and layouts are done.

Anyway, on the DAC boards we decided to use OpAmps as voltage regulators for the analog circuitry. Why you ask? Well, what comprises a good voltage regulator? It should have:

- high PSRR
- low output impedance
- high open loop gain for the feedback to work
- it should behave symmetrically on load steps, be they more or less current being demanded
- also THD seems to matter

In my opinion good OpAmps should tick all of the above boxes. So, on to the actual implementation. I have attached a part of the schemtaic, the AVCC PSU. We will use the OPA1622 because of its low output impedance compared to other devices in the same family. The output stage being symmetric should help with both symmetric behaviour on load changes and it should also halve the output impedance vs. just one pass/shunt transistor.

What will have to be determined is if the ferrite bead in the AVCC output line can stay, it will probably have to be replaced by a 0 ohm resistor simply because the lowest impedance beads have a too high impedance even at audio frequencies.

What I'm wondering about is the usage of a snubber at the output of the opamp. The output of the opamp behaves as an inductance at higher frequencies which will produce overshoot and ringing in the small signal step response. So i thought the a snubber there would be beneficial. From a paper by TI i got that the snubber capacitance should roughly be three times the "parasitic" capacitance, which in this consists of 2x10u + 2x4u7, so i chose it to be 100u. The resistor value is calculated as sqrt(L/C) (it's saying 1R for the moment, please ignore the actual value for now). So the big question is how to find out the output inductance of the opamp.

I have a theory as to how to calculate that but I'm not 100% sure. Maybe somebody can shine some light on that. Also I'm looking forward to your opinion if that snubber makes any sense at all in this position.

Looking forward to your answers and stay healthy!

sgrossklass 22nd March 2020 02:12 AM

Can't say I'm terribly convinced.

Noise wise, the LT3042 would be in the same ballpark as the OPA1622 already. If you're going to use an RC to feed the opamp, I'd rather make that 1k / 100, maybe even lower in impedance.

I would also consider giving "princess on a pea" type components some RC filtering of their own, or possible opt for designs with some half-decent PSRR already (so maybe not your average CMOS inverter oscillators).

Before resorting to an overdesigned contraption like this, I'd try to come up with the best possible implementation of the bare LT3042 first (C702 positioning etc.; consult datasheet).

I mean, I have nothing against this old opamp as voltage regulator trick per se (a good way of supplying fussy oscillators), but the economics of this design just don't strike me as very good. The LT3042 is almost entirely going to waste, as is the second half of the OPA1622. If you wanted high-end, you'd be using a 5 V reference and an ultra-low-Vnoise part (AD797 class). If you wanted cheap, you'd be using an 78M09/12, an R/RC divider and I dunno, an NJM4580 or a 5532 perhaps (or an OPA1688)? Something that isn't too fussed by capacitive loads in any case.

Do you have any estimate on required output current? Frying away (+18-5) V = 13 V, the thing might get a bit toasty.

Vovk Z 22nd March 2020 08:14 AM

Doesn't output opamp have to be connected as integrator?
Large capacity on the opamp output and its stability. Can it work ok with such a circuistry?

jimk04 22nd March 2020 09:32 AM

Why not just use the ESS datasheet opamp based psu for AVCC. I think they used an LTC6655 as 3v3 ref.

I am no expert here, just have read a lot on Mark's es9038q2m thread

Tfive 22nd March 2020 10:54 AM

Quote:

Originally Posted by sgrossklass (https://www.diyaudio.com/forums/power-supplies/351471-using-opamps-voltage-regulators-post6129750.html#post6129750)
Can't say I'm terribly convinced.

Noise wise, the LT3042 would be in the same ballpark as the OPA1622 already. If you're going to use an RC to feed the opamp, I'd rather make that 1k / 100, maybe even lower in impedance.

I would also consider giving "princess on a pea" type components some RC filtering of their own, or possible opt for designs with some half-decent PSRR already (so maybe not your average CMOS inverter oscillators).

Before resorting to an overdesigned contraption like this, I'd try to come up with the best possible implementation of the bare LT3042 first (C702 positioning etc.; consult datasheet).

I mean, I have nothing against this old opamp as voltage regulator trick per se (a good way of supplying fussy oscillators), but the economics of this design just don't strike me as very good. The LT3042 is almost entirely going to waste, as is the second half of the OPA1622. If you wanted high-end, you'd be using a 5 V reference and an ultra-low-Vnoise part (AD797 class). If you wanted cheap, you'd be using an 78M09/12, an R/RC divider and I dunno, an NJM4580 or a 5532 perhaps (or an OPA1688)? Something that isn't too fussed by capacitive loads in any case.

Do you have any estimate on required output current? Frying away (+18-5) V = 13 V, the thing might get a bit toasty.

OK, one by one:

I chose 10k/10u because there are decent ceramic caps available in this value, 100u might become pretty big and costly... and the opamp sees an impedance of <1k above 15.9Hz anyways, so the resistance part of the RC is not that dominant for overall impedance.

What the heck is a princess of pea device? Why are you talking about oscillators? This is the analog reference supply for the DAC. You probably got it confused with VCCA... There we will use LT3042 standalone, as for the DVCC.

The power dissipation will be below 200mW which is a piece of cacke for the OPA1622 with its thermal pad (AVCC only draws around 2mA quiescent and around 8mA output current).

How do you determince which opamp is (too) fussed about capacitive loads? How would the OPA1688 fir the bill in a better way (or the other mentioned opamps). I will stay the heck away from AD797, as it's prone to instability and I definitely don't want to risk that.

Tfive 22nd March 2020 10:55 AM

Quote:

Originally Posted by Vovk Z (https://www.diyaudio.com/forums/power-supplies/351471-using-opamps-voltage-regulators-post6129907.html#post6129907)
Doesn't output opamp have to be connected as integrator?
Large capacity on the opamp output and its stability. Can it work ok with such a circuistry?

Why an integrator? The Opamp is configured in gain = +1 config here, it essentially mirrors/buffers the input voltage to its output pin.

From what I read such large capacities don't pose a risk of making the opamp unstable. Phase margin is still maintained.

Tfive 22nd March 2020 10:59 AM

Quote:

Originally Posted by jimk04 (https://www.diyaudio.com/forums/power-supplies/351471-using-opamps-voltage-regulators-post6129939.html#post6129939)
Why not just use the ESS datasheet opamp based psu for AVCC. I think they used an LTC6655 as 3v3 ref.

I am no expert here, just have read a lot on Mark's es9038q2m thread

The ESS datasheet does not have any schematics afaik. You probably mean the application note by ESS which uses AD797 as buffer and no extra reference, the just use DVCC and a 10k/10u RC low pass.

Why don't we plan to use the LTC6655?
a) it's not cheap and we need three references on the board
b) LT3042 is already in the bom and performs very similarly
c) with an RC low pass after the reference I doubt there will be a lot of noise difference after the RC.

Tfive 22nd March 2020 11:02 AM

Btw. everything that I say here is to be taken with a grain of salt, I' no expert either, everything self taught about electronics.

Still would be looking forward to get the inductance/snubber question out of the way though...

MorbidFractal 22nd March 2020 11:57 AM

Quote:

Originally Posted by Tfive (https://www.diyaudio.com/forums/power-supplies/351471-using-opamps-voltage-regulators-post6129992.html#post6129992)
From what I read such large capacities don't pose a risk of making the opamp unstable. Phase margin is still maintained.

Don't know where you read that.

http://www.ti.com/lit/ds/symlink/opa1622.pdf

High Capacitive-Load Drive Capability: > 600 pF

Does not translate to slapping 10uF on its output. If you care to sample the data sheets of a wide range of operational amplifiers you will find that most if not all recommend that when driving capacitative loads that load is isolated from the feedback node by a low value resistor, 22R or greater.

https://dl.khadas.com/Hardware/ToneB...0Datasheet.pdf

Makes little to no reference to dynamic loading of AVVC but just gives a value of 6mA. Without knowing more I would be inclined to be surprised if the DAC core is operated in a way that introduces excess spurious current draw. Rather than asking here drop an e-mail to the manufacturer.

Tfive 22nd March 2020 12:15 PM

Quote:

Originally Posted by MorbidFractal (https://www.diyaudio.com/forums/power-supplies/351471-using-opamps-voltage-regulators-post6130037.html#post6130037)
Don't know where you read that.

http://www.ti.com/lit/ds/symlink/opa1622.pdf

High Capacitive-Load Drive Capability: > 600 pF

Does not translate to slapping 10uF on its output. If you care to sample the data sheets of a wide range of operational amplifiers you will find that most if not all recommend that when driving capacitative loads that load is isolated from the feedback node by a low value resistor, 22R or greater.

Will try to find that reference and post it here. But given that these Opamps are used frequently with such high capacitors on their output I'm not really worried about that. After all the capacitance is 4-5 orders of magnitude greater than the mentioned 600pF (which is really beasty compared to the 50pF of an 1611 for ex.)

Quote:

Originally Posted by MorbidFractal;6130037
[URL="https://dl.khadas.com/Hardware/ToneBoard/Datasheet/ES-ES9038Q2M-DAC%20Datasheet.pdf"
https://dl.khadas.com/Hardware/ToneBoard/Datasheet/ES-ES9038Q2M-DAC%20Datasheet.pdf[/URL]

Makes little to no reference to dynamic loading of AVVC but just gives a value of 6mA. Without knowing more I would be inclined to be surprised if the DAC core is operated in a way that introduces excess spurious current draw. Rather than asking here drop an e-mail to the manufacturer.

Well. the current can swing from 0 to 7.8mA if operating into a virtual ground in mono mode. See datasheet page 52, to the end of the page.


All times are GMT. The time now is 07:07 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 17.65%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2020 DragonByte Technologies Ltd.
Copyright ©1999-2020 diyAudio

Wiki