Capcitor type for Vref

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
...
The best approach is to add external Very Low Noise and ultra low impedance Vref Generator. Below is a circuit I often use. Pls. note, it’s for designs where Vref is 2.5V. Due to the “High” impedance of the internal Vref (VBias) Gen, they don’t normally mind the addition of the external Gen. Nevertheless; it’s prudent to match the output voltage of the External Generator, as close to the output voltage of the internal Gen as possible.

Due to the Feedback of the OP-AMP, the impedance of the Generator appears almost “flat” within the Audio band (in fact, the output impedance is much LOWER at LF) – thereby limiting the sonic “footprint” of the VRef Capacitor used. Its important to realise that RF decoupling is still required, and the OP-Amp will offer very little correction - due to its limited gain Bandwidth at RF. The RF quality of the 100nF ceramic is critical, try placing it directly across the DAC Vref Pin and Ground – and the electrolytic C5 should be a Low ESR such as an OS-Con or Tant. The quality will now be less critical, due the active correction of the OP-AMP.

Note, the feedback is taken FROM the DAC Vref pin; this removes the effect of PCB wiring and any noise. The limitation at LF is now due to the DAC’s package bonding impendence.

The FB connection should be as short as possible back to the Op-Amp.

19746d1073832412-capcitor-type-vref-dac-vref-gen-jpg

I'd like to try that approach on a WM8741 DAC. But it is not quite clear to me how to connect to such:

WM8741_vmid.gif
 
I’d check first to make sure the voltage at pin 11 and 18 is half of AVDD.

Remove C9 and C10 (they are most likely 10uF). Solder 0.1uF SMD capacitors in their place – you can get the npo’s at this capacitance but you’ll need to look hard. Make sure they are properly soldered at both ends so that they won’t be susceptible to any mechanical resonance; place them as close to the IC as possible. Note: these capacitors are actually a part of a Vref Generator (C6).

Then, fire-up the Vref Generators and ensure their output is the same as “half of AVDD” – see the first sentence above. The schematic posted by JohnW is for a Vref voltage tuned to 2.5V value.

Feed the 5534 out (from a first Vref Generator) to pin 11. Via a SEPARATE wire, connect the same pin 11 to 5534 INV input. I’d use short, solid core pure copper wires for this. Do not just bridge 5534's OUT and INV pins!!

Repeat for pin 18 with a second Vref Generator.

Download the WM8741 pdf – you’ll understand the pin numbering (and “remove C9 / C10”) I used above.
 
I’d check first to make sure the voltage at pin 11 and 18 is half of AVDD.
Sure.
Remove C9 and C10 (they are most likely 10uF). Solder 0.1uF SMD capacitors in their place – you can get the npo’s at this capacitance but you’ll need to look hard. Make sure they are properly soldered at both ends so that they won’t be susceptible to any mechanical resonance; place them as close to the IC as possible. Note: these capacitors are actually a part of a Vref Generator (C6).
Understood.
Then, fire-up the Vref Generators and ensure their output is the same as “half of AVDD” – see the first sentence above. The schematic posted by JohnW is for a Vref voltage tuned to 2.5V value.
I'm thinking of feeding AVDD with a opamp-buffered voltage reference: "The MAX6072 is a dual-output precision series voltage reference. The product features two outputs, +VREF and +VREF/2. The device exhibits a very low 1/f noise of 2ppm (peak-to-peak)." In that case +VREF/2 was the reference for AVDD/2.
Feed the 5534 out (from a first Vref Generator) to pin 11. Via a SEPARATE wire, connect the same pin 11 to 5534 INV input. I’d use short, solid core pure copper wires for this. Do not just bridge 5534's OUT and INV pins!!
Understood, that would make the whole attempt useless.
Repeat for pin 18 with a second Vref Generator.
Why certainly!
Download the WM8741 pdf – you’ll understand the pin numbering (and “remove C9 / C10”) I used above.
I know it inside and out already.

Thank you very much! :worship:
 
Voltage Reference for WM8741

The VGen idea is very interesting and I intend to use such voltage reference in my DAC. But I want to use less noisy circuit. So, after few days of searching proper solution of the voltage reference for WM8741 I found Texas Instruments' chip: REF6225.
It has:
- Temperature Drift: 3 ppm/°C (max) from 0°C to +70°C
- Output Noise: 5 μVRMS with 47-μF Capacitor
- Integrated Drive Buffer
– Output Impedance: < 50 mΩ (0-200 kHz)
- Supply Current: 820 μA
- Initial Accuracy: ±0.05%
- Output Current Drive: ±4 mA

I have just drawn a schematic of the VRef. My DAC will have a dual mono setup so I use four such voltage references.
 

Attachments

  • M5600 WM8741 DAC V-Refs Left.PDF
    125.4 KB · Views: 152
Hi zenelectro

Thank you for interesting and replay.

I use REF6225 Texas Instruments because the IC is High-Precision Voltage Reference included driver buffer in outputs. It has one important feature: nearly flat output impedance less than 50mOhms in range 100Hz - 1MHz with single small capacitor. Look at the diagram of output voltage drop comparison, too.

Suggested LT3042 has lower output noise but unknown output impedance.
 

Attachments

  • REF6225 driving REF pin of ADC.JPG
    REF6225 driving REF pin of ADC.JPG
    92.9 KB · Views: 311
I read LT3045 datasheet but cannot find output impedance value. I choose REF6225 because I have not spare rooms near the DAC chip. My idea is maximal condensed of parts and short traces between them. So 3-4 LT3045s and few RsCs around each regulator will take more room on PCB and make traces longer than using single REF6225.

But your suggestion is very interesting. Maybe I will use the LT3045 to supply VCXO clock.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.