I was hoping to get some feedback on a design I am working on.
I would like to make a transformerless, IC based direct injection box for stereo use, that is powered off of the phantom supply coming from the mixer. This DI will be used for sending mp3, drum machine, click track, etc samples to the mixing board. I am basing my design off of Rod Elliott's mono implementation:
ESP - Direct Injection Box for Recording and PA Systems
I am assuming I can use two of these circuits in one aluminum box. The stereo connection will be a single 3.5mm TRS jack, insulated from the box. The left and right circuits are both powered by their respective left and right XLR phantom supplies. Each XLR pin 1 will be connected to the box. The tip of the jack will go to one circuit, the ring of the jack will go to the other circuit, and the sleeve will go to both circuits as the audio ground.
Am I overlooking anything? Is the grounding scheme sound? Is this a good way to go about sending a balanced stereo signal without using transformers and using phantom powering?
Thanks for any feedback.
I would like to make a transformerless, IC based direct injection box for stereo use, that is powered off of the phantom supply coming from the mixer. This DI will be used for sending mp3, drum machine, click track, etc samples to the mixing board. I am basing my design off of Rod Elliott's mono implementation:
ESP - Direct Injection Box for Recording and PA Systems
I am assuming I can use two of these circuits in one aluminum box. The stereo connection will be a single 3.5mm TRS jack, insulated from the box. The left and right circuits are both powered by their respective left and right XLR phantom supplies. Each XLR pin 1 will be connected to the box. The tip of the jack will go to one circuit, the ring of the jack will go to the other circuit, and the sleeve will go to both circuits as the audio ground.
Am I overlooking anything? Is the grounding scheme sound? Is this a good way to go about sending a balanced stereo signal without using transformers and using phantom powering?
Thanks for any feedback.
In this case it seems advisable to connect the case to star ground via one 10r/100n combo. I would also try to look out for a better low-power audio opamp than the TL072 - TLE2072 maybe? There ought to be some decent low-Ib bipolar types, too (NJM4560/4565?). Use a resistor matching R3 in the feedback of U1A for better impedance balance.
As shown, power supply rejection for the virtual ground is mainly provided by the zener regulator. As P48V shouldn't be super dirty to begin with, this should be generally adequate, but could be much improved by a minor change to supply filtering - replace C4 by a 220µ (25V min, 35V recommended) and move its negative terminal from virtual ground to real ground, i.e. in parallel to D3. A 10µ ought to be plenty for C5 then.
BTW, you can get away with a far simpler solution as long as you are willing to make one set of cables for each value of source output impedance - or for each range of 100 ohms or so at least, which should still yield decent CMRR in practice. You simply follow the recommendations of Jensen Transformers' Bill Whitlock, which goes as follows:
Use a shielded twisted pair cable with usual XLR pinout at the other end. You may choose to leave XLR pin 1 unconnected if the input suffers from "pin 1 problem", or use a 10R-1k in parallel with 10n or so.
At the unbalanced end:
Connect output to hot.
Connect ground to cold via a series resistor that matches source output impedance as precisely as possible. Best low-frequency CMRR is obtained if any output capacitor is also included (yay for DC-coupled outputs).
Connect ground to shield.
(Phantom power obviously needs to be turned off.)
Pros and cons of this solution:
+ Electrically simple
+ High CMRR and as such immunity to interference can be achieved
+ No power needed
- Does not increase levels (while the above circuit has a gain of 2)
- Does not provide symmetric waveform for hot and cold that would allow input stage distortion to cancel
As shown, power supply rejection for the virtual ground is mainly provided by the zener regulator. As P48V shouldn't be super dirty to begin with, this should be generally adequate, but could be much improved by a minor change to supply filtering - replace C4 by a 220µ (25V min, 35V recommended) and move its negative terminal from virtual ground to real ground, i.e. in parallel to D3. A 10µ ought to be plenty for C5 then.
BTW, you can get away with a far simpler solution as long as you are willing to make one set of cables for each value of source output impedance - or for each range of 100 ohms or so at least, which should still yield decent CMRR in practice. You simply follow the recommendations of Jensen Transformers' Bill Whitlock, which goes as follows:
Use a shielded twisted pair cable with usual XLR pinout at the other end. You may choose to leave XLR pin 1 unconnected if the input suffers from "pin 1 problem", or use a 10R-1k in parallel with 10n or so.
At the unbalanced end:
Connect output to hot.
Connect ground to cold via a series resistor that matches source output impedance as precisely as possible. Best low-frequency CMRR is obtained if any output capacitor is also included (yay for DC-coupled outputs).
Connect ground to shield.
(Phantom power obviously needs to be turned off.)
Pros and cons of this solution:
+ Electrically simple
+ High CMRR and as such immunity to interference can be achieved
+ No power needed
- Does not increase levels (while the above circuit has a gain of 2)
- Does not provide symmetric waveform for hot and cold that would allow input stage distortion to cancel
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.