DI Design tradeoffs -Gain Vs. Distortion

[Renfield]

I'm not sure I want to know.... :p

 

Sometimes it's best not to ask.

[agedhorse]

A couple of comments that might help clear up some things here...

 

A transformer used for a passive DI MUST drop the level because of the very basic operation of a transformer which is to transform signal impedances. So as a basic design parameter, say we need to have an output impedance to drive the line of 10x the input impedance of a typical console (~1.5k) means the transformer will have a secondary impedance of 150 ohms.

 

Now the input of a passive transformer isolated DI should be around 20-50k, so the IMPEDANCE ratio is going to be 50k/150 ohms = 333. that means the winding ratio for a transformer like this will be 18:1 (the impedance varies as the square of te winding ratio) or about a 25dB drop in level to go with impedance transformation.

 

In practice, we find that many folks will use a lower ration because the REAL input impedance is not based sole on the transformer but on the load that the transformer is driving (the load impedance is reflected back through the transformer) so this works to our advantage by making the transformer operate more lightly loaded. So in practice, if you have 1.5k loading the secondary, and you want 50k input impedance on the primary, then the real (or practical) transformer IMPEDANCE ratio would be closer to 33 (or maybe even 100 for losses) and the winding ration would be 10:1 and the dB loss would be ~20dB. Then if you choose to make the input inmpedance 25k, the winding ratio would decrease even more. 6:1 up to 10:1 is pretty common.

 

Now for an active DI, the driving (source) impedance is pretty low so let's say it's 200 ohms, a winding ratio of 2:1 would allow the box to drive a load 4 times lowere than it would with a 1:1 ratio.

 

Generally, it's best to target a DI (of any type) to match the middle of the mic gain range, or about -30dB level (average of -50 to -10), so if the active stage driving the transformer puts out -10dBu or whatever, you will want to look at a loss of around 20dB or 10:1.

 

A DI that can deliver +4dB or greater can be a big problem on some sound systems, especially those without input pads.

 

Note that these examples are nominal levels and they will vary all over the place. We have to choose some form of common sense reference whatever it may be.

[knuckle_head]

...... We have to choose some form of common sense reference whatever it may be.

 

Crap - what is this common sense of which you speak?

 

 

 

 

 

 

 

 

 

 

 

My head isn't fully around all this yet, but reading this and other similar things helps. L. Ron is so so sick of me.....

 

[agedhorse]

Crap - what is this common sense of which you speak?

 

The common sense is that you have to balance all of the design elements so that you get a signal that is suitable in level for the majority of combinations inputs and outputs that the unit is likely to be used with. It's tougher than you might casually think unfortunately.

 

Then there's the tradeoffs of transformer design... higher turns ratios have other issues that must be overcome (just costs more money) and then there's termination issues (not very important for bass applications) and shielding etc.

[knuckle_head]

Right - light bulb moment when the transformer ration/squaring-of-value thing was explained to me.

 

I am working through stepping up from a low impedance balanced signal - stands a few things on its head.