Are you really fine with IC voltage regulators ?

Mark, the capacitor is used for DC decoupling- not AC signal coupling. Most here would never use a class II dielectric for the latter function. Film and electrolytic caps are severely limited at HF because of ESL/ESR - the very things you say X7R are bad at.
Put the lack of understanding together with the prevailing audiophile dogma and you get people who replace the supposedly crappy 10uF X7R (not to mention X5R) LT304x output cap with a huge audiophile-approved through-hole film cap and then blame LT304x for the poor result :rofl:
 
  • Like
Reactions: 1 users
What I'd really want is a lt3042 for grown up men. Say TO-220-5

A while ago, Jan and I did discuss whether a DIY version can be made.

The idea is to allow more flexibilities, whether higher voltage (100's of volts) or higher current (2~5A or so).
It turns out to be more difficult than I thought, but then I am not fit with opamps anyhow.
So maybe the guru's here have better ideas.

The LT304x essentially consists of 2 parts :
1) a voltage reference consisting of a stable current source, a resistor for converting current to voltage, and a cap to suppress the noise;
2) an opamp in follower mode to drive the output load.

The input voltage of the LT304x is limited to 20V.
In theory one can float its "Gnd" to allow much higher voltage.
But the voltage at the opamp +Vin has to be restrained at start up, or the opamp will die.
Increase in current can be done with external PNP or NPN, as in 3042 datasheet P.27.
But can we not just do it with a high-power opamp, or an opamp plus a FET, e.g. ?

The current source is not so difficult.
A long tail JFET plus a source resistor will get us a 10~100µA CCS which is probably stable and low noise enough.
But getting a unity-gain stable, high current opamp to remain so with a high capacitive load is not always simple.

The other features we don't always need.
At least I don't, other than current limit, at times.



Patrick
 

indra1

Member
2010-11-05 6:44 am
Bogor
:blush: I must be getting too old for this. I totally forgot that I had changed the AVCC setup in my ES9038Q2M. LT3042 is the regulator but it does not feed AVCC. The following op amp buffer feeds both AVCC and I/V reference. No wonder it sounds so good in subjective sighted listening tests :) But in my AK4490 design LT3042 feeds the Vref directly. That sounds even better :rofl:
I belive anybody claiming hearing differences between non pathological DAC implementation (and all Toppings products are definitely not) is not a discussion serious partner, and qualifies as a delusional audiphool.,,,
@bohrok2610
Since we all know that you are no audiophool, there is a good chance your implementation of ES9038Q2M or both is somewhat pathological.
 
A while ago, Jan and I did discuss whether a DIY version can be made.

The idea is to allow more flexibilities, whether higher voltage (100's of volts) or higher current (2~5A or so).
It turns out to be more difficult than I thought, but then I am not fit with opamps anyhow.
So maybe the guru's here have better ideas.

The LT304x essentially consists of 2 parts :
1) a voltage reference consisting of a stable current source, a resistor for converting current to voltage, and a cap to suppress the noise;
2) an opamp in follower mode to drive the output load.

The input voltage of the LT304x is limited to 20V.
In theory one can float its "Gnd" to allow much higher voltage.
But the voltage at the opamp +Vin has to be restrained at start up, or the opamp will die.
Increase in current can be done with external PNP or NPN, as in 3042 datasheet P.27.
But can we not just do it with a high-power opamp, or an opamp plus a FET, e.g. ?

The current source is not so difficult.
A long tail JFET plus a source resistor will get us a 10~100µA CCS which is probably stable and low noise enough.
But getting a unity-gain stable, high current opamp to remain so with a high capacitive load is not always simple.

The other features we don't always need.
At least I don't, other than current limit, at times.



Patrick
Patrick, you basically describe my T-reg regulator :cool:
Using a current source and R to set the (filtered) ref, a floating opamp error amp and a pass element.
It works from 5V or so to 500V by changing just 1 resistor ...

Jan
 
But not 2~5A ?
Or you already tested all combinations ? :)

BTW, the current source you use is seemingly not particularly low noise.


Patrick
400mA @ 500V with the 12N60xxx.
A hefty FET gives you 5A at up to, ohh, 100V?
Choose your pass device by SOA, set the current limit accordingly, and the sky is the limit!
And the ref noise can be made almost arbitrarily low.

Jan
 
You can stack instead of amplify PTAT voltages, that's the way the bandgap from the 1960's Electronics Letters article works. That's much less noisy.
For the record, it was Hilbiber's article in the 1964 ISSCC Digest of Technical Papers rather than Electronics Letters. Apparently this was the very first article about bandgap references. In the early 1970's, variants that needed much less voltage and produced much more noise were developped, and those are used everywhere now.