My take on a discrete shunt voltage regulator

[ikoflexer]

Originally Posted by AndrewT

latest version takes you to post 261 where the same link appears?

The second line of my signature takes me to post # 232

[stormsonic]

[QUOTE=ikoflexer]I will still use the 2N5087s that I have. I haven't tried BC560+BC560, but I have tried 2N5087+2N5087, and still the better one is to use MPSH81 on the bottom.[QUOTE]

I found. Maybe I need new reading glasses

[keantoken]

I've tried in vain to find RF PNP transistors that could match the NPN equivalents.

Here's my current preferences for audio purposes:

Very Fast, low-gain: 2N5769/5771
Higher gain, still pretty fast: 2N4124/4126

If speed is not an issue I use the standards like the BC5xx series. I don't know what's possible with japanese transistors. There seems to be a paradox between Cjo and Tf. A transistor with 4pF Cjo may be rated 100MHz while one with 18pF may be rated 500MHz. So I suspect Cjo is a more important factor here...

- keantoken

[ikoflexer]

Thanks kt. Now you see my problem I could turn the circuit all the way around but that would require p-channel mosfets and that makes me less happy.

There are a few 2saXYXY parts that I've been meaning to try instead of the mpsh81, but it's just an idea at the moment. Even as it is, this regulator is very good I think. But there's always room for improvement.

The other area we can improve is line rejection, but I'm very tempted to do that via capacitor multiplier/gyrator before the reg. Paul Hynes (here going by the maximus user name) has suggested in another thread to somebody else (the Twisted Pear people) to use a cascode in the CCS. It's possible that in other circuits the cascode in the CCS provides a lot of improvement, but in this one here... I find it does not. The cascode in the CCS was one of the first things I tried. So, there...

A few to consider:

2sa1434 pnp high hfe 100MHz
2sa1697F pnp 300MHz hfe=160-320
2sa1749F pnp 400MHz hfe=160-320
2sa1875F pnp -- || --
2sc3820 pnp 200MHz high hfe=800-3200

[keantoken]

How about 2SA3243?

http://www.toshiba.com/taec/componen...c//66/7771.pdf

Have you taken a look at what cascoding does to the line rejection of a series (BJT) capacitance multiplier? Somewhere around -108db I think. However there is an alternative... We can increase OLG in the CCS.

Try this. Homemodder has used this kind in his amplifiers (in low power places), except with BJT output. J1 at the bottom can be made a resistor... This design is really limited only by the current source used and the MOSFET's Cjo. Even a simple resistor gives good performance. (sorry, PSRR graph was taken with a perfect 2mA CCS, the actual graph bottoms at -120db with the current Jfet CCS)

- keantoken

[ikoflexer]

I got similar results without the ideal CCS. I've tried a lot of ideas in the CCS, cap multipliers, and gyrators. Just haven't posted them.

I decided to take time now to produce a proper PCB for v2, and stay away from other distractions. It's mostly finished, but I've been pulled away from it by various other things. It'll look something like this, still have to bring it up to date with release 5c. It's going to be 2.76'' x 2'' so it's not too large, but not miniature either.

[Terry Demol]

Originally Posted by ikoflexer

Thanks kt. Now you see my problem I could turn the circuit all the way around but that would require p-channel mosfets and that makes me less happy.

There are a few 2saXYXY parts that I've been meaning to try instead of the mpsh81, but it's just an idea at the moment. Even as it is, this regulator is very good I think. But there's always room for improvement.

The other area we can improve is line rejection, but I'm very tempted to do that via capacitor multiplier/gyrator before the reg. Paul Hynes (here going by the maximus user name) has suggested in another thread to somebody else (the Twisted Pear people) to use a cascode in the CCS. It's possible that in other circuits the cascode in the CCS provides a lot of improvement, but in this one here... I find it does not. The cascode in the CCS was one of the first things I tried. So, there...

A few to consider:

2sa1434 pnp high hfe 100MHz
2sa1697F pnp 300MHz hfe=160-320
2sa1749F pnp 400MHz hfe=160-320
2sa1875F pnp -- || --
2sc3820 pnp 200MHz high hfe=800-3200

Ixo,

These are worth a look. There are a whole series which include compliments.
The specs are very good for the right application.

Sanyo Pico Tr series


cheers

T

[ikoflexer]

Ah, thank you Terry! Hm, surface mount. Not an obstacle, but it does change some of my plans a bit. If only I could find a good source for these; so far digikey and mouser say no such thing.

[ikoflexer]

An update and a bit of good news. I've finished the pbc layout and built another prototype tonight using the board. The board is made around rev 5c posted before. I aimed for about 20V output, and fed it 45V DC because that's what I had around. The board has room for two LEDs, but I used just one because the voltage across it was about 2.1V. The following values I used in the real circuit:

R2 = 2K9
R7 = 470R
R9 = R11 = 15K

With these the output voltage was about 20V, and the current limit was about 500mA.

[stormsonic]

and 2N2819 (on PCB) => 2N3819