Negative voltage linear regulator for a preamp

Hitec

Member
2010-01-19 1:08 pm
I am designing a USB DAC using CP2114 + PCM1794A and I need dual supplies for the current-to-voltage opamp output stages. I try to design the circuit to work from USB 5V supply so I have a switching regulator module that produces +-12V from 5V and I want to post regulate that linearly to +-9V. +-5V or something else could also work fine for max. 2Vrms output so let's not constraint the requirements too much.

The problem is now the negative side linear regulator. There is not much discussion or examples for these in the internet. Good PSRR and low noise are important figures I think. Suggestions?
 
You're in luck, needing to lose 3 V from the positive 12 supply. You can use a 2.4 v zener from the +12 to the base of a pnp transistor, hooked up as an emiter follower. Load goes on the emiter. Don't forget the e-caps. Calculate your current load to determine of you need a transistor with a tab, and a heat sink. Calculate the resistor for the zener to keep the power of the zener below ~60% of the rating. You can't use a darlington transistor with this, zeners don't come lower than 2.4 v. If you need enough current to need a 1.2 v drop darlington transistor, you'll have to stack up 1n4148's diodes to get 1.8 v drop on the base reference. don't forget to limit current into the base lead with a resistor.
I'm thinking of running a LM1875 from 80 v no center tap transformer. This has 60v supply max. Making the +12 v rail from the plus supply is simple, just use a LM34012 or 7812. making the 68 v rail from the +80 is not quite so cookbook, I have the national linear voltage regulator handbook from 1975, which hqs the LM340-12 in it (that's minus 12) which is their equivalent to the Fairchild 7912. the application they have shows two matched reguoators working off a center tapped transformer. So, useless tutorial, I'm hoping I could attache the "ground" pin of the 7912 to the +80 rail, but I don't know. I can't experiment right now because newark didn't have any 7912 last time I ordered. Orders are $7 a box so I delay until I have several things I need at once.
 

Hitec

Member
2010-01-19 1:08 pm
You're in luck, needing to lose 3 V from the positive 12 supply. You can use a 2.4 v zener from the +12 to the base of a pnp transistor, hooked up as an emiter follower. Load goes on the emiter. Don't forget the e-caps. Calculate your current load to determine of you need a transistor with a tab, and a heat sink. Calculate the resistor for the zener to keep the power of the zener below ~60% of the rating. You can't use a darlington transistor with this, zeners don't come lower than 2.4 v. If you need enough current to need a 1.2 v drop darlington transistor, you'll have to stack up 1n4148's diodes to get 1.8 v drop on the base reference. don't forget to limit current into the base lead with a resistor.
I'm thinking of running a LM1875 from 80 v no center tap transformer. This has 60v supply max. Making the +12 v rail from the plus supply is simple, just use a LM34012 or 7812. making the 68 v rail from the +80 is not quite so cookbook, I have the national linear voltage regulator handbook from 1975, which hqs the LM340-12 in it (that's minus 12) which is their equivalent to the Fairchild 7912. the application they have shows two matched reguoators working off a center tapped transformer. So, useless tutorial, I'm hoping I could attache the "ground" pin of the 7912 to the +80 rail, but I don't know. I can't experiment right now because newark didn't have any 7912 last time I ordered. Orders are $7 a box so I delay until I have several things I need at once.
I understood nothing :confused:
 

Hitec

Member
2010-01-19 1:08 pm
I am actually having hard time evaluating do I even need a dual supply. So what happens if I connect those opamp positive inputs to some voltage, say 1V, instead of grounding them in the figure 31. above? I guess this would shift the output up so that I could get rid of using a negative supply. I will put caps at the output stage so I can have an artificial ground there. It is just that all application examples have a dual supply of +/-15V or so.
 
I am actually having hard time evaluating do I even need a dual supply. So what happens if I connect those opamp positive inputs to some voltage, say 1V, instead of grounding them in the figure 31. above? I guess this would shift the output up so that I could get rid of using a negative supply. I will put caps at the output stage so I can have an artificial ground there. It is just that all application examples have a dual supply of +/-15V or so.

Thta's a good point. The DAC output is centered around 2.5VDC. The normal way to process that is with an opamp with a 2.5V 'ground'. And the opamp then can be a single supply 5V one.

Jan
 

Hitec

Member
2010-01-19 1:08 pm
Thta's a good point. The DAC output is centered around 2.5VDC. The normal way to process that is with an opamp with a 2.5V 'ground'. And the opamp then can be a single supply 5V one.

Jan
Ok are you sure about that? I mean the datasheet does not give a lot of information on those outputs. It is just confusing that why so many designs seem to be using dual supplies even though it is more complicated. AFAIK almost all designs block DC from the outputs anyway.
 
...ok well the datasheet example applications do not block DC which is even more confusing.

Of course. If the 'opamp ground' is 2.5V, and the signal is centered around 2.5V, the output is centered around zero V. No blocking needed.
But in that case the opamp must have a neg supply because the signal output is around zero.

Jan
 

00940

Member
2003-09-03 12:17 pm
Liège
Careful guys... the pcm1794's output must be kept at 0V (or near). There are protection diodes at the output and if you bias them, you raise the distortion. So you need a bipolar supply if you want opamps to do the I/V conversion.

Of course, since there's a current offset, you have a voltage offset at the output of the I/V opamps; but since it's common to the + and -, you get rid of it with the following opamp.

What's your switching module? Does it have a regulated output ? Since you have +/-12v, a 2vrms output and opamps, tight regulation really isn't a big deal. Cleaning up the HF noise from the switcher is the main point. A LC filter followed by a cap multiplier would be a good solution in your case.
 

Hitec

Member
2010-01-19 1:08 pm
Careful guys... the pcm1794's output must be kept at 0V (or near). There are protection diodes at the output and if you bias them, you raise the distortion. So you need a bipolar supply if you want opamps to do the I/V conversion. --

What's your switching module? Does it have a regulated output ? Since you have +/-12v, a 2vrms output and opamps, tight regulation really isn't a big deal. Cleaning up the HF noise from the switcher is the main point. A LC filter followed by a cap multiplier would be a good solution in your case.
Well that 0V is what I feared it will be.

I ordered this one (TMH 0512D) as the dual supply and it is not regulated. However, I am in a design phase so anything might change :) I considered the regulated switcher + LC-filtering too and maybe I will implement it.

What is a cap multiplier?
 

00940

Member
2003-09-03 12:17 pm
Liège
For the cap multiplier : Capacitance Multiplier Power Supply Filter In that case, it uses a big electrolytic and a big transistor. In your case a small pnp and a mkt film cap around 3.3uf are a better choice. The reason behind it: Using 3-pin regulators off-piste: part 2

A solution to have a single rail I/V would be to use a passive I/V (a simple resistor, something like 47r) followed by a coupling cap and a differential to single ended stage biased to V+/2 (with gain to get to 1 or 2vrms).