Building a symmetrical PSU B1 buffer

diyAudio Chief Moderator
Joined 2002
Paid Member
Changing the CCS current changes the Id of both the big mosFETs.

Transconductance of mosFETs depends very much on Id.
MosFETs run at low Id do not amplify well (due to low transconductance).

That's why I was surprised to see irfp240 being specified for the first DCB1.

It was asking the wrong devices to do the two jobs being asked of them.
That is why we now see/hear performance changing when we change currents.
We are moving the devices onto different parameters/characteristics.

The various Salas Shunt Regulators that followed the DCB1 were designed to perform with better layouts and better devices.

See gfs VS Id charts for IRFP9240 and IRF9530. Both from Intersil. I am not that good into reading them spot on due to lack of finer gradation and different x axis scaling. Looking roughly equal at say 1A? They did not appear changing Zo much in LTSPICE if interchanged at same bias currents in same regs non the less. (9530 is used in SSLV1.1).
 

Attachments

  • gfs9240.gif
    gfs9240.gif
    11.2 KB · Views: 396
  • gfs9530.gif
    gfs9530.gif
    12.9 KB · Views: 394
diyAudio Chief Moderator
Joined 2002
Paid Member
What is the extra current actually achieving ?

The B1 only consumes the Idgs current of the 2SJ170 which is approx 10mA.

The B1 is pretty immune to PSU noise so it cant be down to PSU noise.

I'm not criticizing, I believe there may well be a sonic difference. I'm just inviting comments as to why.

I've got a standard B1 with a low impedance power supply and it sounds wonderful. I'll use any comments to see how my DCB1 compares with my B1.

I was going to rip the B1 apart for components for the DCB1 but Im going to keep them both to compare them. Other than the output caps.

Has nothing to do with consumption. Has to do with the regulators. DCB1 is regulators 90% and a textbook SE CCSed JFET buffer. Thanks to Nelson Pass it was reminded that such a buffer it can be all there is we need in many cases. All I did was to treat it royally so it will give its 100%. I have given options on boards for setting anything according to each one's aspirations. Can run from no sinks to mega sinks. I originally specified TO-247s so to run without sinks on their own heat absorbing mass reliably at about 60mA per rail, and to have easy sturdy mounting due to insulated eyelet for heavy use if chosen so. IRP9240/240 are reliability dogs, another reason for their choice, especially in a DC circuit where if one rail goes, you get the other one's DCV smack on output. You will see many with no sinks and 12-0-12VAC Tx in the beginnings. People gradually voted heat by majority, its a trend. You experience, you decide. Its flexible. Just make it first.
 
OK - Time for a bit of drilling.

Time to go trawling through these thousands of posts to find a few notes about the LEDs.

So far I've measured GREEN = 2.1V
YELLOW = 1.9V
RED = 1.65V

Is it just important to get approximately 9V and 5v4 as long as + and - are the same ?
 

Attachments

  • 001.jpg
    001.jpg
    218.4 KB · Views: 343
anywhere close to those 9v and 5v4 is quite good enough.

You will find that for each colour that different models of LED will show a big variation in Vf at your chosen If.
You will also find that within a single LED type from the one manufacturer there is a very small change in Vf at your chosen If.

You can fine tune If, to bring the sigma(Vf) to your desired total.
 
Measuring LED voltages

I've been busy measuring my collection of LEDs - well the 3mm RED and GREEN ones.

I reckon that the best match that I can get for 9V and 5V4 is as follows.

9V = RED RED RED GREEN GREEN. (1.61+1.62+1.61+2.08+2.08 = 9V0)

5V4 = RED RED GREEN. (1.63+1.63+2.11 = 5V37)

I can then match both strings to within 10mV of each other and the target voltage.

Is that sufficiently accurate ?
 

Attachments

  • DSIR9337.JPG
    DSIR9337.JPG
    199.5 KB · Views: 320
diyAudio Chief Moderator
Joined 2002
Paid Member
Looking at POST 138 you are using traffic lights to acurately produce the reference voltages - Am I right in this assumption ?

You mean the pictures of the black original boards in post 138 I guess. Yes some combine colors. Try 9.4V total, the old ones had an additional drop of a 10R that boosted the 9V a bit. No sinks low Id build that one. Easy to squeeze in crowded box.
 

Attachments

  • MezGru.jpg
    MezGru.jpg
    169.2 KB · Views: 405
diyAudio Chief Moderator
Joined 2002
Paid Member
I've been busy measuring my collection of LEDs - well the 3mm RED and GREEN ones.
I reckon that the best match that I can get for 9V and 5V4 is as follows.
9V = RED RED RED GREEN GREEN. (1.61+1.62+1.61+2.08+2.08 = 9V0)
5V4 = RED RED GREEN. (1.63+1.63+2.11 = 5V37)
I can then match both strings to within 10mV of each other and the target voltage.
Is that sufficiently accurate ?

How did you measure those? This is the best method to measure near real circuit operating conditions:
 

Attachments

  • MeasLED.gif
    MeasLED.gif
    20.6 KB · Views: 391
Hey there, I am just starting to source components and wanted to double check the transformer selection before proceeding. Is an Antek AN-0515 appropriate? Specs read 15V and 50VA, but I can't tell if it is 15+0+15 or not (or if I should be using a different product altogether). Link is here .

Or does anyone know of a more logical transformer choice (lower the price, the better; for someone in USA).