I am looking into a balanced to single ended converter for the PGP amp. I was originally thinking about a simple solution using LM4562 but then I said perhaps 2.7nV/rtHz is pretty high... and the slew rate is at no match with the PGP amp... So I decided to design and experiment a discrete devices converter that now measures pretty good (matching the overall performanace of the PGP amp), although it's rather complicated and quite expensive to build (2SK170/2SK74 input stage, 2SK246/2SJ103 current sources, a total of about 20 trannies and a handful of passives) and, worst of all, it requires a large PCB with little space in the case to properly shield against electric and magnetic fields. It can be done but it would be expensive and overall difficult.
Then I have noticed the new TI/BB OPA211. This critter is stunning: 1.1nV/rtHz, 80nVpp noise 0.1Hz to 10Hz, THD+N: -136dB (G = 1, f = 1kHz) and -120dB at 20KHz at 600ohm load and 3V output (I need only 1V and the load is over 10k), 0.35mV offset, 27V/uS SR (this could be better, though...) 45MHz BW at G=1, unity gain stable, 3.6mA supply current, CMRR and PSRR 120dB at 20KHz, etc... all for a little over $7. The data sheet specifies "professional audio preamplifiers" among intended applications. I had a short call with the TI guys and I was told that this device performance secret is a new SiGe process. A dual version OPA2211 is on it's way. They are also working on a FET input opamp (OPA827) in the same SiGe process. They'll send me a couple of OPA827 evaluation samples.
Have any of you experienced with this OPA211 device? Do you think it's worth further bothering with a discrete design for the balanced to single ended converter? I know, this OPA211 is not even built out of dirty sand, it's dirty sand polluted with some dirty germanium, so it cannot sound good, but from this perspective I'm willing to gamble.
Datasheet: http://focus.ti.com/lit/ds/symlink/opa211.pdf
Then I have noticed the new TI/BB OPA211. This critter is stunning: 1.1nV/rtHz, 80nVpp noise 0.1Hz to 10Hz, THD+N: -136dB (G = 1, f = 1kHz) and -120dB at 20KHz at 600ohm load and 3V output (I need only 1V and the load is over 10k), 0.35mV offset, 27V/uS SR (this could be better, though...) 45MHz BW at G=1, unity gain stable, 3.6mA supply current, CMRR and PSRR 120dB at 20KHz, etc... all for a little over $7. The data sheet specifies "professional audio preamplifiers" among intended applications. I had a short call with the TI guys and I was told that this device performance secret is a new SiGe process. A dual version OPA2211 is on it's way. They are also working on a FET input opamp (OPA827) in the same SiGe process. They'll send me a couple of OPA827 evaluation samples.
Have any of you experienced with this OPA211 device? Do you think it's worth further bothering with a discrete design for the balanced to single ended converter? I know, this OPA211 is not even built out of dirty sand, it's dirty sand polluted with some dirty germanium, so it cannot sound good, but from this perspective I'm willing to gamble.
Datasheet: http://focus.ti.com/lit/ds/symlink/opa211.pdf
Did you try THS4131 for this application? It is quite old but it will perform balanced to unbalanced conversion (and vise versa), it's voltage noise is 1.3 nV/rtHz and current noise is half of what OPA211 has. THD for 1V output into 10K could be below 120 dB across the audio range - at the time I've used it I did not have means to measure below 110 dB.
Cheers
Alex
P.S. - later TI released "audio" version of the same chip with somewhat changed front-end - OPA1632. I still prefer THS4131.
Cheers
Alex
P.S. - later TI released "audio" version of the same chip with somewhat changed front-end - OPA1632. I still prefer THS4131.
x-pro said:
Alex,
I have experimented extensively with THS4131/OPA1632 and I agree it's an amazing piece of silicon. I wouldn't bother to use anything else if in need of a diff in/diff out amp.
The problem is that the excellent performance is available only for the differential output (and that's mostly because of the distortion cancelling). The single ended performance (that is, output taken from any of the two output pins, referred to the ground) is at best mediocre.
A single ended output is exactly what I am looking for (and with a gain of 1). One option is to use a discrete opamp in a differential amp configuration, optimizing/trimming for CMRR. This is what I originally did, with good measured results. The only problem is the size of the board required to accomodate two channels, the costs, and the difficulties (mostly due to the size as well) to shield against magnetic fields.
Then I realized that a few OPA211 could, based on the data sheet, potentially deliver the same performance as a discrete version, at about 1/10 the size and price. My question was if anybody has any experience with these new opamps and if there are any (theoretical or practical) known issues.
syn08 said:
The poor single ended THD is mostly due to how the nfb is applied in this kind of opamp configuration (differentially) - one output is referenced to the other, not ground. In otherwords, the opamps nfb only works to make an amplified replica of the input signal between it's two outputs.
G.Kleinschmidt said:
Could be, but I'd rather think this IC was really not designed to handle single ended output. It's intrinsic distortions are relatively high, however symmetrical distortions tend to cancel even-order harmonics.
See this article http://focus.ti.com/lit/an/slyt165/slyt165.pdf
syn08 said:
Yes, I've got that article and have designed several discrete opamps using the outlined topology.
You are right that the IC in question isn't designed for single ended output, but that is the whole point - any fully differential opamp will provide mediocre performance if an output is taken in a single ended fashion. There is no signal reference to ground in such a topology and that is why the additional servo is required to steer the common mode voltage at the outputs.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.