The AD820 has enormous offset (1mV) and LF noise (2µV p-p from 0.1-10Hz). IMHO, it is not a good choice for a DC servo. The OPA140 and ADA4625 are better, but the Analog ADA4522 has an even better combination of low noise, low input bias, low offset and drift, low cost, and low distortion. I'm using it in a servo circuit and it's really nice. It's a real 'chopper' amplifier (zero offset), so it has essentially no 1/f noise and very low DC offset and drift, parameters very important for a DC servo. When used as an integrator, the chopping noise artifacts are filtered out completely. In the circuit I'm using, the residual offset is typically on the order of 4-8µV, a very low value. Plus, no extra noise is injected into the main circuit, using only a resistive post attenuator and no RC post-filter.
Last edited:
You're not running the part open loop, right? That's the only case where differential input resistance matters - when you're running the amplifier as a comparator or some other application where linear feedback is essentially defeated. For example, if you ground the - input and drive the + input with your signal, running the amplifier as a comparator, the + input will behave like a 30kΩ resistor since there is no feedback at all.
When you operate the amplifier in its linear region with feedback, which is what happens with a typical DC servo integrator circuit, the voltage across the 30kΩ differential input (i.e. the voltage across the + and - input terminals of the amplifier) will be the output voltage divided by the loop gain, which is around 140dB at DC. So, the 30kΩ will essentially be magnified by 140dB or 10^7, resulting in an extremely high "input impedance".
The penalty to using a JFET and not a 'zero drift' CMOS amp is that the JFET amplifier will exhibit lots of offset and 1/f noise, as well as a sensitivity to mechanical flexure of the die, all of which are removed by the zero drift (chopper) function of a modern zero drift CMOS amp.
The tradition for DC servos in audio circuits, which has persisted into some of the current literature, is to use relatively shoddy JFET amps with poor voltage offsets and 1/f noise, but these days we have chopper amps that can eliminate many significant problems associated with old JFET amp designs. IMHO, there's no reason to take this poor performance, especially when using them as an integrator, a circuit design that completely eliminates the chopper artifacts, and allows the excellent DC performance and low drift, as well as 'zero' 1/f noise, of a chopper stabilized amplifier.
Practically, this week I have been poking at a handful of circuits using the ADA4522-2 as a DC servo for a balanced amplifier, and they not only simulate well, the PC boards and prototypes behave just as well. My Keysight 34461A meter has a hard time measuring the residual DC offset of the composite amplifier, but I estimate it to be around 4-8µV, roughly in line with what the simulation predicts. So, these parts actually do work as excellent DC servos in simulation and reality.
When you operate the amplifier in its linear region with feedback, which is what happens with a typical DC servo integrator circuit, the voltage across the 30kΩ differential input (i.e. the voltage across the + and - input terminals of the amplifier) will be the output voltage divided by the loop gain, which is around 140dB at DC. So, the 30kΩ will essentially be magnified by 140dB or 10^7, resulting in an extremely high "input impedance".
The penalty to using a JFET and not a 'zero drift' CMOS amp is that the JFET amplifier will exhibit lots of offset and 1/f noise, as well as a sensitivity to mechanical flexure of the die, all of which are removed by the zero drift (chopper) function of a modern zero drift CMOS amp.
The tradition for DC servos in audio circuits, which has persisted into some of the current literature, is to use relatively shoddy JFET amps with poor voltage offsets and 1/f noise, but these days we have chopper amps that can eliminate many significant problems associated with old JFET amp designs. IMHO, there's no reason to take this poor performance, especially when using them as an integrator, a circuit design that completely eliminates the chopper artifacts, and allows the excellent DC performance and low drift, as well as 'zero' 1/f noise, of a chopper stabilized amplifier.
Practically, this week I have been poking at a handful of circuits using the ADA4522-2 as a DC servo for a balanced amplifier, and they not only simulate well, the PC boards and prototypes behave just as well. My Keysight 34461A meter has a hard time measuring the residual DC offset of the composite amplifier, but I estimate it to be around 4-8µV, roughly in line with what the simulation predicts. So, these parts actually do work as excellent DC servos in simulation and reality.
Just to be clear, the OPA140 is not a 'shoddy' JFET amplifier - it performs extremely well for a non chopper stabilized amplifier. However, the voltage offset is ~20x higher than the ADA4522, and it will exhibit 2x the noise in the 0.1-10Hz band, a region where a DC servo circuit will have significant gain. The OPA140 will exhibit much lower input bias current, which is an indication of the somewhat high input switching artifacts of the chopper design of the ADA4522. However, with careful choice of input resistor values, one can still achieve excellent residual offsets with the ADA4522. My 'end of the day' servo performance using the ADA4522 has ~10x lower offset than what could be expected from an OPA140 at lower cost.
My complaint is with somewhat recent literature that still suggests that JFET amplifiers with very high offsets and 1/f noise should be used as a DC servo integrator, when there is no good reason to do that anymore. Zero drift (chopper) amplifiers do have some strange quirks, but they also do some amazing things, such as completely removing 1/f noise and providing extremely low drift and residual offset, with almost no penalties in an integrator application such as a DC servo.
My complaint is with somewhat recent literature that still suggests that JFET amplifiers with very high offsets and 1/f noise should be used as a DC servo integrator, when there is no good reason to do that anymore. Zero drift (chopper) amplifiers do have some strange quirks, but they also do some amazing things, such as completely removing 1/f noise and providing extremely low drift and residual offset, with almost no penalties in an integrator application such as a DC servo.
The ADA4522-1 is the single version, and it uses the standard single op amp pinout, without any enable or offset trim pins. The AD820 also uses the standard single pinout, but pins 1 and 5 are used for offset null, so if your circuit uses those, disable the offset control circuitry.
The ADA4522-1 is only available in surface mount packages, so if your PCB uses DIP packages, you need to use an adapter board to fit the ADA4522-1 into your existing circuit. TI sells a nice kit of adapter PCBs for low $: DIP-ADAPTER-EVM DIP Adapter Evaluation Module | TI.com
Thank you, yes are also for my Erno's mono 75W power amp, all IC of above link are the same?
http://es.farnell.com/w/search?range=inc-in-stock&st=ADA4522&numero-de-pines=8pines
TIA
OPA189 is an opamp with outstanding data.
- high precision few uV offset
- low noise 5.2 nV
- high GBW 14 MHz
- high slewrate 20 V/uS
- low THD 0.00006 %
- high output current +/-65mA
Draws little current 1.3mA typ
Supply voltage ±2.25 V to ±18 V, Absolute Max +/-20V
Unity Gain Stable
Doesn't cost very much
Comes in SOIC-8
Last edited:
The OPA 189 and AD4522 are simliar in the key parameters
Inital offset (typical)
189 - 0.4uv
4522 - 0.7uv
0.1 to 10 hz noise
189 - 100 nv
4522 - 117 nv
Price as of July 17
189 - $2.04. TI direct 10 pieces
4522 - $2.42 - Mouser 10 pieces
So with the specs slightly better and the price 20 % lower the 189 looks like a winner, but ( shipping costs maybe a bigger issue than the chip prices themselves, unless your buying other parts at the same time)
Inital offset (typical)
189 - 0.4uv
4522 - 0.7uv
0.1 to 10 hz noise
189 - 100 nv
4522 - 117 nv
Price as of July 17
189 - $2.04. TI direct 10 pieces
4522 - $2.42 - Mouser 10 pieces
So with the specs slightly better and the price 20 % lower the 189 looks like a winner, but ( shipping costs maybe a bigger issue than the chip prices themselves, unless your buying other parts at the same time)
I checked my spreadsheet of DC servo op amps, and the OPA189 is really good, a toss up in some ways compared to an ADA4522. The extra bandwidth and better distortion may make it cleaner in some servo circuits, but I lowpass the send to the servo amp, so this isn't so important for me. The only 'worse' spec is the input bias spec, which is 2x the value of the ADA4522, and that will increase the end of the day residual offset from the servo. But, in my circuit, the increase is from 4µV to 5µV, both ridiculously small values, well inside of the unit to unit variation between actual ICs.
The show stopper for me is that I need a dual IC for the servo, to make the two chopper clocks synchronized. This is the OPA2189, but it's still not shipping yet. Two singles is not the same, and is potentially dangerous if there is any clock leakage - two different clocks could beat against each other and cause in-band birdies. I can't take that chance, so I need to stick with a dual if I want to use a zero drift amp in this balanced amp circuit. But, if your servo is a single, then the OPA189 sounds like a nice choice.
The show stopper for me is that I need a dual IC for the servo, to make the two chopper clocks synchronized. This is the OPA2189, but it's still not shipping yet. Two singles is not the same, and is potentially dangerous if there is any clock leakage - two different clocks could beat against each other and cause in-band birdies. I can't take that chance, so I need to stick with a dual if I want to use a zero drift amp in this balanced amp circuit. But, if your servo is a single, then the OPA189 sounds like a nice choice.
The OPA2189 is not shipping but you can get samples from TI. I just put 2 in my Audio-gd DAC19 as replacements for the OPA2134PA that where in there as servo's. The sample ones are labeled as prototypes but because they are about to be on the market it is safe to say that they are likely to be in their final stage.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.