PP logo


Phantom powered direct inject box.



A direct inject box ( DI-box ) box provides the impedance and level matching required to connect a high impedance unbalanced music instrument signal or high level signal to a mixing-console's balanced microphone input.
The most common input connector is a ¼" TS ( tip-sleeve or mono) jack, in most cases in parallel with a second ¼" TS jack to pass signal on.
The input should have a high input impedance in order not to load the signal source and accept a wide range of signal levels from instrument pick-ups, electronic synthesizers, drum-machines, loudspeakers, etc.
The output is on a 3-pin male XLR connector.

The DI-box described here is a one- or two-channel active, phantom powered unit in a relatively compact enclosure.
It runs off a P-48 phantom supply.
The only "feature" in this design is a 2-position sensitivity switch.
I would have liked a ground-lift function, but this is almost impossible to implement without using a transformer, so it was left out. If you REALLY need it, you will find that the ground-lift implementation in most active DI-boxes is insufficient anyway.
I have left most other special functions ( HP-filter, polarity inverter, equalizer ( cable simulator ) ) out as I find these belong in the FOH mixer where you can hear their influence on the final mix.

A very important part of a good DI-box design is a proper enclosure so switches and connectors can not be bent or broken even when the unit is exposed to some serious abuse.
The standard enclosure I used for this design is not optimal, but the best I could find.
I have included a description of how to build a proper enclosure ( some mechanical work is required ).
You may find that the electromechanical components I suggest for this design are on the expensive side, but I like to build things to last. If you don't, feel free to use cheaper parts.


DI05 block schematic.
Fig.1: DI05 block schematic.

The Di05 board contains one complete amplifier channel and 2 output circuits to allow the construction of a single-channel unit using 1 board or a compact 2-channel unit using 2 boards.
The blocks in the above schematic do what their names say.
"Output Matrix" is a fancy name for some jumpers to wire 2 boards together ( it was easy to design, not so easy to document ).
The details are shown in Appendix A.

2-channel unit.
Fig.2: 2-channel unit.

The upper drawing shows the connections between 2 boards where the distance between the front and rear panels is 100 mm ( as the board is designed for ) and the lower drawing for a different panel distance.
The board is split into 3 sections that can be cut apart in 2 marked locations.
The minimum building height is 32.5 mm, but you will most likely want a little more to make room for the jack-plugs.
I have used the lower board for channel 1 and the upper board for channel 2. If you want it differently, swap a few wires ( do not mail me about the "right way" to do it ).

Input Connectors, HF filter, Attenuator and Amplifier schematic.
Fig.3: Input Connectors, HF filter, Attenuator and Amplifier schematic.

J11 and J12 are standard ¼" jack connectors. These are shown as stereo jacks ( I prefer these as they have more solderings to hold the PCB and connector together ), but are connected as mono jacks.
They must be connected as mono jacks so they do not short the signal during plug insertion or removal.
The jack connectors are types with the sleeve terminal connected to the chassis for best HF immunity.
The "IN" connector is the signal source and the "LINK" pass the signal to the next unit in the chain.
R14 shunts the input to ground to reduce noise ( a little ) if no input connector is plugged in.
The jacks can be used in series with a speaker provided the power is limited ( approximately 60 W/8 Ω, 30 W/4 Ω ).
If you need higher power, use jacks with a higher current rating ( they will not fit in the boards ).
The maximum input voltage before clipping is 35 V, corresponding to 150 W/8 Ω, 300 W/4 Ω.

R11 and C11 is the input HF filter. The cut-off frequency is around 700 kHz with low source impedances, falling to around 70 kHz with a 10 kΩ source impedance.
The input impedance is 470 kΩ//220 pF ( 0 dB ) or 430 kΩ//220 pF ( -20 dB ).
For 470 kΩ//220 pF at both gains, R12=422 kΩ and R13=52.3 kΩ.
A 220 pF input capacitance is approximately the same as 2 m of instrument-amplifier cable.
My original design from 1992 used 15 pF for C11, but that was before cell-phones, Wi-Fi, EMI standards and similar nasty things.
C12 can be used to reduce the upper frequency in the -20 dB position ( 390 pF will give -3 dB at 10 kHz ).
The -20 dB setting of the attenuator is the "general" setting and is suitable for most sources like instrument pick-ups or loudspeakers.
If your application requires that you have signal on the input without phantom voltage present, you need to use the -20 dB position to avoid that the input clamp your signal through its protection diodes.
The 0 dB position is used for "electronic" sources with a limited output voltage, typically synthesizers, drum-machines or DJ-mixers.

U11 is an amplifier with a current consumption of 2 mA or less per amplifier. The board will accept DIL-8 or SOIC-8 cases.
Suggested types are LM833 ( all testing is done with this ), TL072 ( higher noise ) or OPA1678. The OPA1678 may be interesting as it is more tolerant to EMI than the other two types.
The OP-AMP must be unity-gain stable and have a reasonable slew-rate ( 5 V/µs..10 V/µs ). Do not use high-speed OP-AMPs - the board is not designed for them.
The asymmetrical bias divider ( R16, R17 ) compensates for the LM833 input bias current.
If you use an OP-AMP with low input bias current, use 220 kΩ for R17.
D13..D16 protects the OP-AMP output from over-voltages due to the charge/discharge of the output capacitors ( fig.4 ).
D11, D12 protects U11A input from over-voltages.

Output connectors.
Fig.4: Output connectors.

The 2 output connector circuits are identical except for the designators.
C16, C17 are the AC coupling capacitors. Their values may seem a little large ( -3 dB frequency is around 1 Hz ), but this is required to maintain good CMRR at low frequencies.
The 63 V rating is because I have a lot of them. 35 V is sufficient.
R22..R25 is a "balanced" output attenuator with a gain of around -15 dB and an output impedance around 200 Ω. "Balanced" as its center-point ( VCC ) is grounded so common-mode signals are attenuated too.
In order to achieve a good impedance balance, R22..R25 should be matched 1% types or 0.1% types.
The attenuator reduce the maximum ( approximately 5 V ) output level to a level that is reasonable for a microphone input and at the same time it reduce the noise from the electronics in DI-box.
C19, C20 is HF decoupling for the microphone-cable. These should be 5% types.
C18 reduce the impedance imbalance introduced by C19 and C20 and create a LP filter around 80 kHz.

Voltage regulator.
Fig.5: Voltage regulator.

A voltage regulator for a design like this is normally just a zener diode and a bypass capacitor, they idea being that the microphone amplifier CMRR will cancel the ( common-mode ) noise from the regulator.
While this worked well in my prototype, around 9 of 10 units I made later had higher output noise than advertised.
The source of the noise was the zener-diode, so I had to measure zener-diodes before use. Around 1 in 10 was useful.
Low voltage zener-diodes ( 5.1 V ) seems to have 20 dB to 40 dB lower noise than 24 V types, but I have only tested a few.
To meet the specification, the noise from the zener and a 100 µF capacitor had to be below -100 dBu.
Adding more capacitance will reduce the noise, but a lot of capacitance is required due to the low impedance of the zener.
This will in turn increase the start-up time of the DI-box, but few people are interested in waiting minutes for the circuit to start up.
The solution is a voltage dependent capacitance multiplier.
When VCC is below the conduction voltage of D19+D20, the only capacitance across the supply is C23. This will charge in <1 s, and you have audio out.
When C21 has charged so D19+D20 conducts, the capacitance multiplier feedback loop is closed and the effective capacitance across VCC increases to around 30 mF.
R27 and C21 is a low-pass filter that effectively filter the noise from D17. The noise on VCC is well below -116 dBu ( the meter reads -116 dBu as it does with shorted input ).
Q11, Q12 is compound darlington coupled transistors.
D20 can be left out ( shorted ) if you do not want an on-indicator.
An interesting side-effect is that it works as overload indicator when the amplifier starves the shunt regulator of current.
Without D20, D19 should be an 18 V or 20 V type.
D21 is normally not used. Discharging an ESD gun directly on D20 may kill Q12. D21 will in most cases prevent that ( a piezo-electric gas-grill igniter is not an ESD gun as some have suggested ).
D22 is over-voltage protection in case someone ESD test the XLR connection. I like to have this as the active shunt regulator is relatively slow and it is a cheap insurance.
A BZX79C30 or BZX679C33 will allow the circuit to be used with 35 V electrolytics.

Component Selection.


For R22..R25, R62..R65 use matched 1% or 0.1% metal film types.
For other resistors, use 1% metal film resistors.

Film capacitors

Use 5% or 10% MKT capacitors.

Electrolytic capacitors

Use 20% general purpose types with minimum 35 V voltage rating ( 25 V for C14 ).

Ceramic capacitors

For values ≤1 nF, use 5% NP0 (COG), 100 V types.
For OP-AMP decoupling, use 50 V X7R types.


Use LM833, OPA1678 or TL072. See text.


Anything with a current rating of around 100 mA, a voltage rating of >40 V and a hfe of >100 will work.


Most 0.5 W types with a 5% voltage tolerance should work.


This is the specification for the DI05A prototypes.
The unit dBu is dB referred to 0.775 V.
More detailed measurements are in the file DI05A_Measurement.ods in the design file download ( including measurements with TL072 and OPA1678 ).
Measurements are with a 40 Ω source impedance and 2 kΩ load.

Table 1: Specification for DI05A.
Input impedance430 kΩ // 220 pF470 kΩ // 220 pF
Maximum input level35 dBu15 dBu
Voltage gain-30 dB-10 dB
Output impedance200 Ω
Frequency range1 Hz-70 kHz1 Hz-80 kHz
THD+N, 1 kHz, 22 Hz..22 kHz BW, 10 dBu input level0.003%0.001%
THD+N, 20  Hz..20 kHz, 10 Hz..80 kHz BW, 10 dBu input level<0.011%<0.018%
IMD, SMPTE, 60 Hz / 7 kHz, 4:1, 10 dBu input level0.004%0.003%
Output noise, 22 Hz..22 kHz, Rs=40 Ω-114 dBu-122 dBu
Output noise, 22 Hz..22 kHz, Input "open"-114 dBu-119 dBu
Board size (length / width / height)98 mm / 59 mm / 24 mm


DI05 PCB photo.
Fig.6: Photo of mounted PCB for a 1-channel DI-Box. The dimension is 98x59 mm.

I have boards available for this project. See the PCBs page.


DI05A design files.

Known Issues / updates.

No known issues.

Appendix A: Board configurations.

The DI05 board can be cut in 3 places to allow for different one- or two-channel applications.
The dimensions of the individual sections can be found in the mechanical drawing in the download.

2-channel unit, 100mm between panels.
Fig.A1: 1-channel unit, 100mm between panels.

2-channel unit, <>100mm between panels.
Fig.A2: 1-channel unit, ≠100mm between panels.
The 2 board sections are connected with 5 wires. For wire lengths >50 mm, wires 1+, 1- should be twisted.

2-channel unit, 100mm between panels.
Fig.A3: 2-channel unit, 100 mm between panels.
The 2 boards are connected with 5 wires. A pin-header can be used for better mechanical stability, but is more difficult to take apart.

2-channel unit, <>100mm between panels.
Fig.A4: 2-channel unit, ≠ 100 mm between panels.
The 2 boards are connected with 10 wires. For wire lengths >50 mm, wires 1+, 1- and 2+, 2- should be twisted.

Cutting the board.

The boards have cutting lines on the silkscreen layer as an aid in cutting.
Use a hacksaw with a replaceable blade to cut the boards.
Do not use your favorite sharp tools - they will be dull after cutting FR-4 ( unless they are designed for it ).
A saw blade for metal normally works well in FR-4. If the blade is new it may be too sharp. Run it once or twice over the edge of one of the pieces of FR-4 you are going to cut off. This will fix the saw.
Cut gently along the line(s) where you want to cut the board. Do not use force - you may pull the traces off the PCB.
If you need to hold the PCB in a wise, use jaw protectors to protect the PCB. 2 pieces of 4..6 mm plywood works well.
A Dremel style tool with a cutting blade should work well in stead of a saw - I have not tried.
After cutting the board, sand the edges using P100 sandpaper. Put the sandpaper on a flat surface and slide the board edge across.
When you have the edges straight, bevel them slightly to avoid burrs.
When you are finished, check for shorts using an ohm-meter on the nearest pads.
Shorts can be removed with a scalpel.

Appendix B: Mechanical construction.

The box used for this design was what I could get and make by the use of ordinary hand-tools.

Photo of DI05 prototype.
Fig.B1: The prototype.

The main issue with this design is that the switch actuators are not protected by the enclosure in any way so you have to take care they are not broken during use and transportation.

The box I used is a "100*66*43mm Black Aluminum Box" [1]. 100 mm is an internal dimension, 66 and 43 mm are external dimensions.
You can find a drawing of the profile in the "Enclosure" drawing in the download ( sorry I am lazy - I haven't drawn the external decoration part of the profile ).
Holes <12 mm were just drilled. The 22 mm holes for the XLRs were made with a "Q-max" style hole punch.
Hole edges that are visible after assembly were painted with a black permanent marker.
A piece of 2 mm thick rubber foam is glued to the bottom of the case. This does not fall off like rubber feet does.

Some issues with this kind of enclosure ( not only this particular one ):
The face plated are normally painted or anodized. Both finishes are non-conductive, so in order to get an electrical connection between the face plates and the profile, you need to remove the surface finish under the screw heads. This can be done with a flat screwdriver or a drill bit.
The screw heads are some universal profile ( they fit nothing ). Best fit for this enclosure is a Philips #1.
The screws in the picture are standard 2.5 mm pozi-drive machine screws.

A better enclosure

DI02 ( my original design ) photo.
Fig.B2: My original DI-box design. The connectors and switches are protected from damage by the enclosure design.

The enclosure is a piece of standard aluminum profile ( 40 mm * 60 mm with 3 mm wall thickness ) with a u-shaped chassis inside.
The profile is cut to length, all edges are beveled and a hole or two for mounting the chassis is drilled.
Surface finish is glass-peening, anodizing and then printing.
This results in a very rugged enclosure that can be stacked in any way you want for transportation.
You can - literally - drive a 3.5 tons car over it without structural damage ( the surface finish does not like it though ).

DI02 chassis photo.
Fig.B3: The chassis from the box above. It is made from A2-steel ( saves a paint-job ).

To fit a PCB with components in both ends, the PCB or the components must be in more than 1 piece.
The Neutrik D-type XLR used here can be taken apart so the board will go into the chassis.
The PCB in the photo is almost transparent as it is 0.5 mm thick to fit inside the box ( the total height inside the chassis is 33 mm ).

DI21 mechanical mock-up.
Fig.B4: A mechanical mock-up for a 2-channel DI-box. This was simply too compact to be useful.
With the jacks installed, you need tools ( or very thin fingers ) to set the switches.

Another way to fit a PCB into a u-shaped chassis is to use spacers as in fig.B4. The XLR spacer goes on the XLRs, the PCB goes into the box, the 2 "U" shaped spacers are put in place and nuts and screws for the connectors are fitted.
The jack sockets used are Cliff Electronics S4 types.
A third way is to cut slots in the chassis and use a face-plate in one or both ends.

It is possible to make the chassis from a standard aluminum profile too. Simply cut off one of the sides.

Considering how easy this enclosure is to make ( assuming you have the tools ) I am surprised that nobody makes them.
As far as I know, you can not buy them anywhere as a standard product.

Appendix C: Special Components.

Some of the footprints on this board are designed for specific components.
The list below show some alternative components. Please check the datasheet carefully to verify that the alternative components will fit before ordering.

Jack connectors:

The board is designed for and tested with Neutrik NRJ6HF-1 connectors
It is a TRS ( stereo ) jack with 3 switches and the sleeve electrically connected to the chassis.
Electrically, only the tip and sleeve connections are used.
The additional pins improve the mechanical strength between connector and PCB ( the connector is the only mechanical support for the PCB in some cases ).

Other types:
Neutrik NRJ series jack sockets.
Cliff Electronic Components ¼" S1 Offset Nose series jack sockets.

XLR connectors

The board is designed for and tested with Neutrik NC3MBH connectors.
Pin 1 and chassis are connected on the PCB, so a type with or without internal pin 1 to chassis connection can be used. The ground contact must be connected to chassis.

Other types:
Neutrik NC3MAH, NC3MAAH.


The board is designed for and tested with Multicomp 1MS1T2B4M7RE switches.
I will strongly suggest that you get a type with gold-plated contacts for a long service-life.

Other types:
C&K Components 7101MD9AV2BE
E-Switch 100AWSP1T2B4M7RE
Alcoswitch ( TE Connectivity ) A101MD9AV2B

Appendix D: Battery Powering.

The only sensible reason, I can see, to use batteries for a DI-box is that it allows you to use a transformer between the DI-box and the FOH mixer to provide galvanic isolation in case of hum caused by a ground-loop.
Batteries cost money, they must be replaced regularly, they must be removed unless used frequently.
An almost worn-out battery can make you miss that one channel is without phantom power during setup. Batteries newer runs out during setup - always during the show.

The DI05 is prepared for battery powering although I have not tested it.
The supply current is around 4 mA per channel, so two 9 V batteries should last for around 100 hours for a single channel or 50 hours for two channels.

There are 2 pads marked "AMP" and "PSU" on the bottom side of the board. Cut the trace between them on the bottom side of the board. This will separate VCC on fig.3 from the rest of the VCC wiring.

Battery powering schematic.
Fig.D1: Battery powering. 2 channels are shown. For 1 channel, ignore references to channel 2.

Fig.D1(a): Battery power only.
For connection to inputs no phantom power only:
Connect C16, C17 with + facing the OP-AMPs. Replace D22 with wire-link and leave the rest of the components on fig.5 out.
If this circuit is connected to phantom power for extended periods of time, C16, C17 may be damaged.
For connection to inputs with or without phantom power:
C16, C17 must be bi-polar types. Replace D22 with a 2.2 kΩ resistor, replace C23 with a 2200 µF/25 V type, leave the rest of the components on fig.5 out. A 2200  capacitor will not fit in the C23 footprint, but it should be possible to mount it on its side on top of the other components.

Fig.D1(b): Phantom or battery power - manual change-over.
C16, C17 must be bi-polar types.
The switch select between phantom power or battery power. It can be a single double-pole or two single-pole type.

Fig.D1(c): Phantom or battery power - automatic change-over.
The circuit will run off phantom power if available.
C16, C17 must be bi-polar types. D91..D94 should be 1 A diodes with at least 40 V rating. 1N4001 will work.You may need to add 100 µF from the AMP1 and AMP2 terminals to GND. You can extend the battery life and headroom very slightly by using schottky diodes ( 1N5819 ). IMO it is not worth the money.
The switch select between phantom power or battery power. It can be a single double-pole or two single-pole type.

The overload indication "feature" of D20 is only available when the circuit runs off phantom power.

Using a normally-open ground contact in the input jack socket to disconnect the battery is possible if you use a battery per channel.

Appendix E: Changing the Gain.

Although the gain selected for this circuit have proven useful for most applications, you may want to change it.
"0 dB" and "-20 dB" in the following refer to the position of the gain selector switch - not the actual gain.
There is -10 dB gain in the output attenuator to match the dynamic range of the DI-box to a typical microphone input, so the actual gains are -10 dB and -30 dB.
The gain values referred to in the following is the gain of the input attenuator alone.

The idea behind the "0 dB" input is to provide i high impedance input with very low noise. This is only possible with a gain of 0 dB.
The gain of "-20 dB" can be selected to suit your application.
If you can accept higher input noise in "0 dB" you can choose both gains freely ( "0 dB" must have higher gain than "-20 dB" ).
One advantage of reducing the "0 dB" gain is that you get a higher resistance in series with input, improving the protection against over-voltages.
The calculations for the input attenuator are shown below. You can also find them in the spreadsheet in the download ( this also include a noise calculation for the attenuator ).

DI05 input attenuator calculation.
Fig.E1: DI05 input attenuator calculation.

Fig.E1(a) is the actual attenuator circuit.
Fig.E1(b,c) where R11 is only used as HF filter and "0 dB" is 0 dB.
Fig.E1(d,e) where R11 is used as part of the attenuator.

Calculations may seem a little complicated - they are not - it is simple ohm's law.
Vin and Iin are as shown on the schematic. V(R11) means the voltage across R11, I(R11) means the current through R11.
The attenuators are designed from a desired gain ( Av ) and input impedance ( Rin ).
In all cases:
Vin = 1 ( you can choose any number except 0 ).
Rin = 4.70e5 ( choose an input impedance suitable for your application ).
Iin = Vin / Rin = 1 / 4.70e5 = 2.13e-6

Fig.E1(b): Switch = "0 dB".
It is obvious from inspection that Vout = Vin and Rin = R15.
Av = 1
Vout = Vin * Av = 1 * 1 = 1
V(R15) = Vout = 1
I(R15) = Iin = 2.13e-6
R15 = V(R15) / I(R15) = 1 / 2.13e-6 = 4.70e5

Fig.E1(c): Switch = "-20 dB".
Av = 0.1
Vout = Vin * Av = 1 * 0.1 = 0.1
V(R12) = Vin - Vout = 1 - 0.1 = 0.9
I(R12) = Iin = 2.13e-6
R12 = V(R12) / I(R12) = 0.9 / 2.13e-6 = 4.23e5
V(R15) = Vout = 0.1
I(R15) = Vout / R15 = 0.1 / 4.70e5 = 2.13e-7
I(R13) = I(R12) - I(R15) = 2.13e-6 - 2.13e-7 = 1.91e-6
V(R13) = Vout = 0.1
R13 = V(R13) / I(R13) = 0.1 / 1.91e-6 = 5.22e4

Fig.E1(d): Switch = "0 dB".
Av = 0.3
Vout = Vin * Av = 1 * 0.3 = 0.3
V(R11) = Vin - Vout = 1 - 0.3 = 0.7
I(R11) = Iin = 2.13e-6
R11 = V(R11) / I(R11) = 0.7 / 2.13e-6 = 3.29e5
V(R15) = Vout = 0.3
I(R15) = I(R11) = 2.13e-6
R15 = V(R15) / I(R15) = 0.3 / 2.13e-6 = 1.41e5

Fig.E1(e): Switch = "-20 dB".
Av = 0.03
Vout = Vin * Av = 1 * 0.03 = 0.03
V(R11) = R11 * Iin = 3.29e5 * 2.13e-6 = 0.7
Va = Vin - V(R11) = 1 - 0.7 = 0.3
V(R12) = Va - Vout = 0.3 - 0.03 = 0.27
I(R12) = Iin = 2.13e-6
R12 = V(R12) / I(R12) = 0.27 / 2.13e-6 = 1.27e5
V(R15) = Vout = 0.03
I(R15) = V(R15) / R15 = 0.03 / 1.41e5 = 2.13e-7
V(R13) = Vout = 0.03
I(R13) = I(R12) - I(R15) = 2.13e-6 - 2.13e-7 = 1.91e-6
R13 = V(R13) / I(R13) = 0.03 / 1.91e-6 = 1.57e4

Appendix F: Increasing the Lower Cut-off Frequency.

The lower cut-off frequency is around 1 Hz to achieve the low noise in the 0 dB gain position.
It can be increased by reducing the value of C13. 15 nF will give a lower cut-off frequency of 20 Hz.
The noise in the 0 dB gain position will be increased by around 2 dB.


[1] Banggood.com 100*66*43mm Black Aluminum Box, ID: 1138847.

Copyright and disclaimer.

Copyright Notice.
This web-page, including but not limited to all text, drawings and photos, is the intellectual property of Poul Petersen, and is Copyright ©.
Reproduction or re-publication by any means whatsoever is strictly prohibited under International Copyright laws.
The author grants the reader the right to use this information for personal use only.
Any commercial use is prohibited without express written authorization from Poul Petersen.

The information is provided on an "as-is" basis and is believed to be correct, however any use of the information is your own responsibility.

This web-site may contain links to web-sites outside Poul Petersen domain ( www.poulpetersen.dk ).
Poul Petersen has no control over and assumes no responsibility for the content of any web-site outside Poul Petersen own domain.

Poul Petersen does not use cookies to "enhance your experience" on this website.

Poul Petersen, C/Faya 14, 35120 Arguineguín, Las Palmas, Spain.
E-mail: diy@poulpetersen.dk

Poul Petersen DIY index

Poul Petersen notes index

Copyright © Poul Petersen 2020. First published: 20200611, Last update: 20200611. Valid HTML!