http://www.ti.com/lit/ds/symlink/pcm1794a.pdf
In the manaul of the DAC 1794a an example of an IV-stage is shown on page 20. The OPs are loaded with 750 and 270 ohms. Isn't that to low? The distortion increases quite below 1 k.
In the manaul of the DAC 1794a an example of an IV-stage is shown on page 20. The OPs are loaded with 750 and 270 ohms. Isn't that to low? The distortion increases quite below 1 k.
They're trying to keep noise as low as possible.
If you want, you can add a small class A buffer inside the feedback loop of the opamps. It keeps the opamps operating in much better conditions.
If you want, you can add a small class A buffer inside the feedback loop of the opamps. It keeps the opamps operating in much better conditions.
I'm thinking of raising the resistances in the 1k range. This increases the noise a bit, but in the range of <-100db it does not matter. Low thd is more important.
Last edited:
Keep in mind that you need enough headroom in your supplies for the -10.1mA max IOUT current for the first stage.
There is also another thing to consider: it's not certain you get much higher thd at 1vrms into 600r than at 2vrms into 1.2k (as a rough example), depending on the opamps.
@Paul: Thanks for the hint. Current is not a problem.
@00940: I am refering to the NE5532 (#1 opamp). 1 or 9 V doesn't matter, problem is the load. All above 600ohm is allowed. The diff between 600 and 1200 is 0.0007% and 0.0005%.
@00940: I am refering to the NE5532 (#1 opamp). 1 or 9 V doesn't matter, problem is the load. All above 600ohm is allowed. The diff between 600 and 1200 is 0.0007% and 0.0005%.
0.0002% is hardly something worthy of worrying.
Current is a problem btw when combined with the I/V resistor. If you push 10mA through a 1K I/V resistor, you already get a 10V signal. If you go even higher, you will clip your opamp. The 750r of the I/V stage is well optimized.
Actually, the first opamp is loaded with 750//560. You could play with the resistors around U3 as we're already at line level but I wouldn't touch too much the I/V resistor.
If I were you, I'd check the datasheet more carefully, p.25.
If the output of U1 swings down to 7.575V with a 750r resistor, what does that tell you ?
If the output of U1 swings down to 7.575V with a 750r resistor, what does that tell you ?
my pcm1794 has 47ohm I/V. That should give a lot of distorsion, but no distorsion at all. Just super sound.
Attachments
Last edited:
Actually, 47r for a passive I/V is ok. You're not loading an opamp and your resistor is small enough to not develop too much voltage at the dac pins. If you were using a larger resistor, the protection diodes inside the pcm1794 might start conducting, creating large distortion.
Thanks I was not aware of that. Raising the resistor to 1.2k*10.1mAp would be possible and is within 15V minus margin.
I'd personally feel a lot more comfortable with 18V rails on that old of an opamp. Can't remember how well it behaves that close to its supply voltage.
Also, any reason you need this much voltage swing? Especially if you have to subsequently attenuate it, it'd be better to lower the transimpedance gain (remember that low feedback impedances are okay when you're not flowing a ton of current!)
Also, any reason you need this much voltage swing? Especially if you have to subsequently attenuate it, it'd be better to lower the transimpedance gain (remember that low feedback impedances are okay when you're not flowing a ton of current!)
I have no idea what sorts of impedances you're thinking of for the differential amplifier, but 1.2K is 1.5x the I/V circuit on page 21, which is swinging 4.5 Vrms. Given most amplifier's input sensitivity is closer to 1-2 Vrms, 4.5 Vrms means you're going to waste a ton to subsequent attenuation and pitch a huge amount of SNR right out the door, magnitudes more than you're gaining by "tricks" here. Unless, of course, you know that your downstream electronics need that kind of voltage swing. You do want to optimize your global gain structure, right?
Remember the LT1028 is a pretty hefty beast and *must* be run at low feedback impedances to get the best out of it (also why they have C3/C4 to keep it stable). Realistically speaking, raising the feedback network impedances to minimize loading is backwards. As Ben (00940) wrote -- swinging >12V across a 1.2K load is very likely greater distortion than swinging 8V across a 800 ohm load, or 6V across a 600 ohm load. Pay closer attention to the power demands on the opamp than the feedback network.
I'd be running the differential amplifier as close to unity as I could, and tweaking the I/V transimpedance until I hit the sweet spot of my amps input sensitivity. Maybe a bit higher transimpedance gain and some attenuation in the differential amplifier, as TI shows. But definitely an output voltage swing that matches the input sensitivity of my amplifier.
Remember the LT1028 is a pretty hefty beast and *must* be run at low feedback impedances to get the best out of it (also why they have C3/C4 to keep it stable). Realistically speaking, raising the feedback network impedances to minimize loading is backwards. As Ben (00940) wrote -- swinging >12V across a 1.2K load is very likely greater distortion than swinging 8V across a 800 ohm load, or 6V across a 600 ohm load. Pay closer attention to the power demands on the opamp than the feedback network.
I'd be running the differential amplifier as close to unity as I could, and tweaking the I/V transimpedance until I hit the sweet spot of my amps input sensitivity. Maybe a bit higher transimpedance gain and some attenuation in the differential amplifier, as TI shows. But definitely an output voltage swing that matches the input sensitivity of my amplifier.
Op amps have lower THD with higher voltage swing than with lower. e.g. 5-9V into 1k is no problem at all.
They sit higher above the noise floor, i.e. measurements are THD+N. I just had a quick look at some of the better measured opamps on TI's site (not the NE5532, I'll have to look at Samuel Groner's work for that), which points to me that a 1.2 k swinging 6 or so volts is limited by the noise of the opamp and swinging 3 V across a 600 ohm load is also limited by the noise of the opamp.
I mean, do what you want, but unless you need all that voltage swing on the output at the end, you're throwing any potential benefit you may realize away across a later volume control.
I mean, do what you want, but unless you need all that voltage swing on the output at the end, you're throwing any potential benefit you may realize away across a later volume control.
- Status
- Not open for further replies.