PCM1794A - IV stage example

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
And how do you think they measured the DAC THD without a I/V ? The THD given for the DAC includes the I/V and filter stage.

All those figures attached are the figures for the DAC + I/V stage. So the I/V stage is at least as good as those figures.
 

Attachments

  • Capture.PNG
    Capture.PNG
    46.6 KB · Views: 322
Last edited:
You need an active stage (low pass) anyway.

I never use a low pass filter after a 1794 dac. I do not care about good measurements but i care about good sound. The less a signal is manipulated the better it sounds. when using tube amps frequency is limited by output transformer.

I hear no difference between 0.1 or 0.002% THD, so stop looking at specs and start listening/building.
 
Last edited:
10K would result into 100 volts. hehe.

I'll make one last attempt to explain:

In the scheme I've been describing, the I/V resistor and the voltage gain stage's feedback resistor are two different components. The I/V resistor is connected directly from the DAC chip output to ground, not to the output of an op-amp. Located there, the I/V resistor passively converts the DAC signal current in to a signal voltage. The resulting signal voltage is then easily amplified via an medium gain voltage amplifier.

With an op-amp, the required feedback network can be of an arbitrarily high resistance, accepting increased noise as the trade-off, along with a few other potential issues. So, the I/V resistor can be low in value, while the separate feedback resistor can simultaneously be high in value. Which relieves the signal gain stage of necessarily driving a low value feedback resistor as might be the case with an op-amp based transimpedance amplifier.
 
Last edited:
Member
Joined 2017
Paid Member
high speed I/V

I haven't experimented passive I/V for pcm179x but has some experience with it for high sample rate DACs(more than 10MHz and more than 10mA current) which are designed for delta-sigma DAC. High-speed DACs usually use transformer interface as manufacturers recommend it. They are not intended to use in LF but for HF at least more than 20kHz. That's why a small SMD transformer is available and has excellent performance.

If you want to use it for LF like audio application, a transformer is impossible. Then you need to choose passive or active. High current(more than 10mA) with low distortion by active I/V is very difficult. In this situation, passive I/V is the only solution which can achieve -120dB THD(this is for measurement purpose not for music). High-speed DACs have relatively wide compliance voltage, e.g., from +1.2v to -0.5v(AD9717). The suggestion of less than 50 ohms(0.4Vpp) with pcm179x is probably an appropriate value. I think pcm179x can work successfully both by active and passive I/V. Generally speaking, passive has better numbers in high OSR like x128.
 
diyralf was saying the contrary (lower thd with higher voltage) but I suspect he meant THD+N and not simply THD.

THD+N improves with higher voltage swing, not THD by itself. If you look at graphs of THD+N vs output voltage (at a constant load), it goes down first because the noise component improves at higher voltages. Then it usually starts to flatten because the distortion component rises as the opamp is asked to provide more power. And it finally rises when the output nears the PS rails and clipping finally sends THD trough the roof.

Ok that make sense. One more question , will I improve THD+N if I make the gain at unity ? Right now I have it at 0.65 and wanted to increase to 0.85 (x input)

I was thinking to change the 1K Rin/650R Rfeedback to 600ohm Rin and 500 Rfeedback (at about 0.83 gain) . Should I expect better numbers ?
 
With those values, you take the output pins of the DAC out of their optimal area of operation. Please do a search on smms73 and pcm1794, he took the pain to do tons of measurements. Optimum thd is achieved with 22r on the outputs. As you go up in value, THD goes up.
 
Why reduncing headroom and amplify again? Makes no sense.

At times this whole thread doesn't make much sense LoL :)

Here are some suggestions:

- To get max DR or least amount of noise, maximise gain in I-V stage (first opamp).
- Use an opamp with very low voltage noise
- Use one that has high current drive and also is fast.

My suggestion is ADA4898 for I-V. It is specced at 40mA / 150 ohm drive capability, 65MHz bandwidth and around 1nV/rt Hz noise (from memory). This is a nice combination for that application.

For the bal -> unbal converter (+ some LPF) choices are wider however ADA4898 will also do well here since it has such good drive capability.

Job done.

T
 
Did a few experiments. The measurement is quite good. The corner frequency is pretty outstanding. Is there a better IV stage in the market?

Freq. resp.: ±0.06 dB (20Hz..20kHz)
Corner freq.: -3 dB (40kHz), -12 dB/octave
Noise: 1.21 μV (-118.3 dB)
THD (1 kHz): 0.0003%
THD (20 kHz): 0.0004%
Out voltage: +6 dBu (1.55V)
BW: 20kHz
 
Last edited:
Hi,

#26, The problem with active virtual ground transimpedance amplifiers is that DAC chips feed them a wideband, fast slewing, fast settling signal current which is difficult to track without dynamic errors.
+ on that ..... and that´s why no-feedback grounded base (or gate) stages are so smart in this application.
Fast enough, verylow-impedance input and high-impedance current output that allows for large voltage swings ... and very low-THD also.

jauu
Calvin
 
Last edited:
Did a few experiments. The measurement is quite good. The corner frequency is pretty outstanding. Is there a better IV stage in the market?

Freq. resp.: ±0.06 dB (20Hz..20kHz)
Corner freq.: -3 dB (40kHz), -12 dB/octave
Noise: 1.21 μV (-118.3 dB)
THD (1 kHz): 0.0003%
THD (20 kHz): 0.0004%
Out voltage: +6 dBu (1.55V)
BW: 20kHz

So what's the schematic giving you those results ? Out of curiosity, what are you using for testing?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.