For more current output, try this:

Feed the LM regulator via 1 ohm. Feed emitter of 2N2955 (PNP!) also. Take PNP Base to other side of the 1E resistor (regulator input) and PNP Emitter to regulator output.
Now, as load increases the reg. develops voltage across the 1E resistor, and at around 0.6V (0.6A through the regulator), PNP turns on. By the time the regulator is taking 0.7A it is developing 0.7V across the 1E resistor (which is dissipating I^2R = 0.49W), and PNP is turned fully on and well and truly taking the rest of the load!
2 components: 2N2955 (and heatsink of course - water-cooled, anyone?) and 1E 1W resistor.
Drop-out voltage = that of the regulator + Vbe(PNP).
The 2N2955 is rated to 10A, and you can add many of them by adding a (1A rated) diode in series with the regulator and 1E0 (compensates for the Vbe of the 2955/s), and 0E1 resistors in the emitters. The 2955s are then sharing the load to 10 * Ireg each, well within rating, and all protected by the regulator.

Boost the Vout? add 1k regulator out to 'ground' (lifted) and resistors / pot to suit then to ground.
This works for *any* LM regulator, just adjust the 'grounding' resistance.
From the datasheet, LM317 is 1.2V out so, given 1k across that, the grounding resistor will be about (desired Vout - 1.2) / (1.2 / 1k).
So, e.g., given regulator LM7812 and 1k across its output, wanting 13.8V out, grounding resistance will be about (13.8-12) / (12 / 1k) = 150E.

Have fun, folks!
73 de Dex ZL2DEX

For more current output, try this:

Feed the LM regulator via 1 ohm. Feed emitter of 2N2955 (PNP!) also. Take PNP Base to other side of the 1E resistor (regulator input) and PNP Emitter to regulator output.
Now, as load increases the reg. develops voltage across the 1E resistor, and at around 0.6V (0.6A through the regulator), PNP turns on. By the time the regulator is taking 0.7A it is developing 0.7V across the 1E resistor (which is dissipating I^2R = 0.49W), and PNP is turned fully on and well and truly taking the rest of the load!
2 components: 2N2955 (and heatsink of course - water-cooled, anyone?) and 1E 1W resistor.
Drop-out voltage = that of the regulator + Vbe(PNP).
The 2N2955 is rated to 10A, and you can add many of them by adding a (1A rated) diode in series with the regulator and 1E0 (compensates for the Vbe of the 2955/s), and 0E1 resistors in the emitters. The 2955s are then sharing the load to 10 * Ireg each, well within rating, and all protected by the regulator.

Boost the Vout? add 1k regulator out to 'ground' (lifted) and resistors / pot to suit then to ground.
This works for *any* LM regulator, just adjust the 'grounding' resistance.

From the datasheet, LM317 is 1.2V out so, given 1k across that, the grounding resistor will be about (desired Vout - 1.2) / (1.2 / 1k).

So, e.g., given regulator LM7812 and 1k across its output, wanting 13.8V out, grounding resistance will be about (13.8-12) / (12 / 1k) = 150E.

Have fun, folks!
73 de Dex ZL2DEX

Raj,

I just had a look on the below mentioned data sheet. Fromt there ist
shuts down if the internal chip temperature reaches 180 deg C, but htis
value is not tested in production.

The thermal resistance chip - case is 5 deg C/W.

Assuming an input voltage of 31 V and output voltage of 18 V , the
difference is 13 V?? times 1.5 A gives 19.5 W of dissipated power.

If You have a heatsink with indefinite good thermal conductance the
tempderature rise will be 19.5 W times 5 deg C/W = 97.5 deg C.

For calculating the overall thermal rise, see e.g.

*

*Henning Weddig*

*DK5LV*


Am 19.10.2017 um 13:13 schrieb David McGough:

Hi Raj,

You may not have had a big enough heat sink for the 317T or another
stability problem if it starts to limit current at 0.5A. My experience
with about 31V input and 18V to 25V output has shown 1.5A output reliably,
with stable voltage regulation. For a typical datasheet, see the
"Figure 8. Current Limit" graph on page 5 of this PDF:



...As with all ancient regular ICs, the 317T has been cloned by everyone
and specs may vary. The exact parts I'm using came from National
Semiconductor many years ago.

Why I originally chose the 317T for my project is that it seemed an almost
perfect match for the IRF-510---meaning keeping the 510 within maximum
specs under all load conditions...Again, with a big enough heat sink on
the 510, too! ....And, over the last 5+ years of continuous beacon
operation, I've found this premise to be true.

Anyhow, there are plenty of regulator choices out there! ...Have fun and
homebrew!

73, David KB4FXC



On Thu, 19 Oct 2017, Raj vu2zap wrote:

i.e LM338 - my choice over a LM317 series. 317 starts limiting current
quite early even with minimum rise in temperature. I don't recommend
it for anything over 0.5A unless the in-out voltage difference is a few
volts!

Raj

At 19/10/2017, you wrote:

If you are going to build a linear regulator to power the drain
of the IRF510, why not use a LM336? It is good to 5A throughput
at those voltages. A smaller one would do for powering the board
directly, even a LM7805 if the voltage drop is not too great.
Don't forget the reverse voltage protection!

john
AD5YE

Hi Raj,

You may not have had a big enough heat sink for the 317T or another
stability problem if it starts to limit current at 0.5A. My experience
with about 31V input and 18V to 25V output has shown 1.5A output reliably,
with stable voltage regulation. For a typical datasheet, see the
"Figure 8. Current Limit" graph on page 5 of this PDF:



...As with all ancient regular ICs, the 317T has been cloned by everyone
and specs may vary. The exact parts I'm using came from National
Semiconductor many years ago.

Why I originally chose the 317T for my project is that it seemed an almost
perfect match for the IRF-510---meaning keeping the 510 within maximum
specs under all load conditions...Again, with a big enough heat sink on
the 510, too! ....And, over the last 5+ years of continuous beacon
operation, I've found this premise to be true.

Anyhow, there are plenty of regulator choices out there! ...Have fun and 
homebrew!

73, David KB4FXC



On Thu, 19 Oct 2017, Raj vu2zap wrote:

i.e LM338 - my choice over a LM317 series. 317 starts limiting current
quite early even with minimum rise in temperature. I don't recommend
it for anything over 0.5A unless the in-out voltage difference is a few
volts!

Raj

At 19/10/2017, you wrote:

If you are going to build a linear regulator to power the drain
of the IRF510, why not use a LM336? It is good to 5A throughput
at those voltages. A smaller one would do for powering the board
directly, even a LM7805 if the voltage drop is not too great.
Don't forget the reverse voltage protection!

john
AD5YE

Note that this fix is not compatible with Allard's instructions for using the rig on CW.

Also, note that there are two PTT bursts.?
Raj's fix is for the carrier burst that transmits out over the air.
There's also a kerchunk you hear in the headphones.

Jerry, KE7ER


On Thu, Oct 19, 2017 at 01:31 am, Raj vu2zap wrote:

The carrier burst on PTT is due to the C122 charging up through D15 or 16 unbalancing it and allowing a leak.

Quick fix is put a 0.68uF cap from the common of D15/16 to the atten pad R39. A track has to be cut.

If your board has a trimpot there then you can tombstone the pot and add the cap.

A 1K resistor across C107 helps.

The leak is almost eliminated.

The carrier burst on PTT is due to the C122 charging up through D15 or 16 unbalancing it and allowing a leak.

Quick fix is put a 0.68uF cap from the common of D15/16 to the atten pad R39. A track has to be cut.
If your board has a trimpot there then you can tombstone the pot and add the cap.

A 1K resistor across C107 helps.

The leak is almost eliminated.

I tried the relay idea Brent, and it did nothing, so don't go hacking your board on the hope this is going to fix your problem. I still see a burst of full carrier for ~ 6 ms immediately after PTT is applied. I see the same result in CW mode, except there's a pause after the burst before the carrier comes on for the first code element is transmitted. The timing is almost precise on this, but I haven't found anything in the code that seems to correlate with it. The 6 ms does correlate with the relay switching speed, but it seems I'm puzzled on this issue again. Either I'm still overlooking something, or I botched the hack on my board.

Sorry about that.

Gary, N3GO