In article ,
cliff wright wrote:

If a search tells you that they are Darlington devices get
rid of them right off!!!
Darlingtons have a minimum collector/emitter voltage of ~1.5 volts or
more so they will get very hot indeed at 30 amps or so.

Cliff-

Are you sure that applies to Astron-type power supplies? The ones I'm
familiar with do not operate with the pass transistors near saturation.

Perhaps a Darlington device is not the best choice for some other
reason, but I would think a pair of 50 Amp devices could handle 15 Amps
each (30 Amps total) with sufficient cooling, if they are rated for at
least 120 watts each. (I assume 22 volts input, 14 volts output, 8 * 15
= 120 Watts per device.)

Fred
K4DII

On Mon, 14 Jul 2008 19:44:02 -0400, Fred McKenzie wrote:

In article ,
cliff wright wrote:

If a search tells you that they are Darlington devices get
rid of them right off!!!
Darlingtons have a minimum collector/emitter voltage of ~1.5 volts or
more so they will get very hot indeed at 30 amps or so.

Cliff-

Are you sure that applies to Astron-type power supplies? The ones I'm
familiar with do not operate with the pass transistors near saturation.

Perhaps a Darlington device is not the best choice for some other
reason, but I would think a pair of 50 Amp devices could handle 15 Amps
each (30 Amps total) with sufficient cooling, if they are rated for at
least 120 watts each. (I assume 22 volts input, 14 volts output, 8 * 15
= 120 Watts per device.)

Fred
K4DII

--------------------------------------------------------------------------------------

Fred is absolutely correct in saying that the pass transistors in most linear
regulators such as those in the Astron supplies do not operate as saturated
elements. Therefore, the higher CE saturation voltage of a darlington has no
bearing in this discussion. On another point, the 120 watts per device that
Fred mentions is at a case temperature of 25 deg, C. Heatsinks of reasonable
size can't dissipate that much power and keep the case temperature at 25C.

I have a few talking points that nobody has touched on directly.

If the regulator was designed to use normal (non-darlington) pass transistors,
and you substitute darlingtons in their place, some very bad things could (will)
happen.

The output of the regulator will likely try to go to its full input voltage rail
(because of the high gain of the darlingtons). The feedback loop in the
regulator will try to bring it back down to the set output voltage. It probably
can't because the gain of the darlingtons is too high, and the output stays
banged against the input rail (22V).

Even if the regulator loop can bring it down, the control loop isn't compensated
correctly for the high gain of the pass elements, and the regulator will
oscillate wildly. Both conditions are catastrophic for a power supply.

Another point that I would like to mention is that of a transistor's SOA or Safe
Operating Area. There are two graphs in the datasheet for the transistors. The
SOA graph shows the safe combinations of collector current and collector-emitter
voltage AT A CASE TEMPERATURE OF 25 DEG C. The power derating graph shows how
much power the transistor can safely dissipate at a given temperature.

You MUST use thesse graphs when designing a pass element. They will help you
choose the appropriate transistor types and will tell you how many transistors
you have to use in the pass element to safely supply the design load current.

In almost all cases, the current through each transistor in the pass element is
limited to much less than the absolute maximum collector current spec. Thus,
although a 2N3771 transistor has an absolute max Ic of 30A, it won't be able to
pass that amount of current at any value of CE voltage. The datasheet gives the
conditions for that current; the higher the collector current, the lower the CE
voltage allowed.

Sorry for being long-winded, but it's not a 2-sentence discussion.

Cheers!!!
==============

Dave M
Remove all of the Xs from the reply address Dave M
Never take a laxative and a sleeping pill at the same time!!

On Mon, 14 Jul 2008 22:20:21 -0400, Dave M
wrote:

On Mon, 14 Jul 2008 19:44:02 -0400, Fred McKenzie wrote:

In article ,
cliff wright wrote:

If a search tells you that they are Darlington devices get
rid of them right off!!!
Darlingtons have a minimum collector/emitter voltage of ~1.5 volts or
more so they will get very hot indeed at 30 amps or so.

Cliff-

Are you sure that applies to Astron-type power supplies? The ones I'm
familiar with do not operate with the pass transistors near saturation.

Are you sure about that ? Have you looked at the capacitor voltage
with an oscilloscope with full nominal current and minimum mains
voltage ? The goal should be that the capacitor voltage just before
the rectifier starts to conduct is only slightly above the output
voltage.


Another point that I would like to mention is that of a transistor's SOA or Safe
Operating Area. There are two graphs in the datasheet for the transistors. The
SOA graph shows the safe combinations of collector current and collector-emitter
voltage AT A CASE TEMPERATURE OF 25 DEG C. The power derating graph shows how
much power the transistor can safely dissipate at a given temperature.

The SOA is important especially in the short circuit current limiting
case, when both the current and voltage across the transistor is high.
A fold-back current limiting solves this problem.

The SOA derating usually starts above 5 V, so in a properly designed
power supply the input-output voltage difference should be less than
that, so in normal operation, the SOA is not very critical.

Paul OH3LWR

happen.

The output of the regulator will likely try to go to its full input voltage rail
(because of the high gain of the darlingtons). The feedback loop in the
regulator will try to bring it back down to the set output voltage. It probably
can't because the gain of the darlingtons is too high, and the output stays
banged against the input rail (22V).


I dont understand what you mean by this. The voltage gain of the
darlington transistors is less than unity in this case since they are
being used as emitter followers.

I am dropping 8.5 volts at 20 amps or 170 watts is being dissipated by
the transistors. This may not be a real problem. I know my commercial
power supply of similar size only provides about 18 volts to the
collector of the pass transistors and this would result in about half
as much power being wasted. What voltage would you normally expect at
the the collector of the pass transistors. I have also thought that
maybe I dont have that much of a real problem since I doubt if the
power supply will ever be put under the same demands as it does
working into my load bank. Also I think I should add two more
transistors in paralell with the two I have. I think this would
distribute the heat better over my heat sink and have less localized
heating.

I dont think the size of my heatsink is a problem, it is probably
overkill. it is a 8in chimney type heat sink( i think that what you
call them) 8x8x10 inches finned inside and out. The transistor body is
mount on the outside withe the wires going down the inside of the
chimney. The end of the chimney is designed to take a muffin fan.

Jimmie

In article
,
wrote:

The voltage gain of the
darlington transistors is less than unity in this case since they are
being used as emitter followers.

Jimmie-

I agree. Being the devil's advocate, a circuit can have parasitic
elements you don't think about, and can oscillate in a different mode
than it normally operates in. If you were using it to power a
transmitter with a poor antenna, there might be some RF energy entering
the base-emitter junction. The energy is rectified and shifts the bias
conditions. DC output voltage is changed, resulting in a change in the
level of the rectified RF, resulting in a change in the DC output
voltage, et cetera. Normally you would disregard this until/unless you
actually observe a problem.

I agree with Paul, that you want the capacitor voltage to be relatively
close to the output voltage. But there is a limit where hum starts to
appear in the output, and regulation is reduced. In the case of the
Astron RS-20 circuit diagram, the main capacitor voltage is listed as
23.0 for no load and 18.0 under load, for 13.8 VDC no load output and
13.7 VDC under load.

I worked on a friend's RS-20. After repair, I ran it for several hours
at 8 Amps load. This is within the continuous power rating, but the
heat sink really got hot!

Fred
K4DII

The voltage gain of darlingtons is less that unity as a emitter
follower. but not the current gain. The base current is like 20 ma.
and the collector is like 20 amps that's a gain of 100. A good
regulator like a LM723 can regulate that easy. If the gain ain't there
in the darlingtons you need a drive transistor.

I use a 0.1 ohm 50watt resistor series with the output, that gives me
an extra 2 volt drop that isn't across the transistor and keeps 40
watts from heating up the output. You still lose 40 watts. It just
goes up in the resistor. The 0.1 ohm 50 watt is only good for like 22
amps.

Put a fan on the heatsink. 170 watts will get hot all the time.
I put a computer power supply fan on my supply, the output stays alot
cooler.
Once the current drops the fan cools it really fast.

You have a high transformer voltage because it's a 50 amp supply and
you are running it at only 20 amps.

On Jul 15, 8:23*am, wrote:
happen.

The output of the regulator will likely try to go to its full input voltage rail
(because of the high gain of the darlingtons). *The feedback loop in the
regulator will try to bring it back down to the set output voltage. *It probably
can't because the gain of the darlingtons is too high, and the output stays
banged against the input rail (22V).

I dont understand what you mean by this. The voltage gain of the
darlington transistors is less than unity in this case since they are
being used as emitter followers.

I am dropping 8.5 volts at 20 amps or 170 watts is being dissipated by
the transistors. This may not be a real problem. I know my commercial
power supply of similar size only provides about 18 volts to the
collector of the pass transistors and this *would result in about half
as much power being wasted. What voltage would you normally expect at
the the collector of the pass transistors. I have also thought that
maybe I dont have that much of a real problem since I doubt if the
power supply will ever be put under the same demands as it does
working into my load bank. Also I think I should add two more
transistors in paralell with the two I have. I think this would
distribute the heat better over my heat sink and have less localized
heating.

I dont think the size of my heatsink is a problem, it is probably
overkill. it is a 8in chimney type heat sink( i think that what you
call them) 8x8x10 inches finned inside and out. The transistor body is
mount on the outside *withe the wires going down the inside of the
chimney. The end of the chimney is designed to take a muffin fan.

Jimmie

raypsi wrote:
The voltage gain of darlingtons is less that unity as a emitter
follower. but not the current gain. The base current is like 20 ma.
and the collector is like 20 amps that's a gain of 100. A good
regulator like a LM723 can regulate that easy. If the gain ain't there
in the darlingtons you need a drive transistor.

Huh, I thought the 723 was a voltage regulator, not current. Im sorry
but what you are saying just doent make sense to me. It is my
understanding that the greater current gain of the Darlington
transistors is just going result in the 723 seeing a lighter load wich
would be godd for the 723. Perhaps that could be a problem if it were
light enough, maybe this would result in the 723 becoming unstable but
this does not appear to ba a problem with light loads, no loads or
heavy loads.
When I first started working on the power supply there was one tiny
problem using the Darlingtons that was more a result of the poor
design of the origonal circuit than anything else. The the origonal
circuit monitored the ouput voltage at the base of the pass
transistors. Because of the 1.6 base to emiter drop of the darlingtons
the output voltage was a little low. After moving the pick off point
to the emitter I no longer had this problem.

I use a 0.1 ohm 50watt resistor series with the output, that gives me
an extra 2 volt drop that isn't across the transistor and keeps 40
watts from heating up the output. You still lose 40 watts. It just
goes up in the resistor. The 0.1 ohm 50 watt is only good for like 22
amps.

Put a fan on the heatsink. 170 watts will get hot all the time.
I put a computer power supply fan on my supply, the output stays alot
cooler.
Once the current drops the fan cools it really fast.

You have a high transformer voltage because it's a 50 amp supply and
you are running it at only 20 amps.

On Jul 15, 8:23�am, wrote:
happen.

The output of the regulator will likely try to go to its full input voltage rail
(because of the high gain of the darlingtons). �The feedback loop in the
regulator will try to bring it back down to the set output voltage. �It probably
can't because the gain of the darlingtons is too high, and the output stays
banged against the input rail (22V).

I dont understand what you mean by this. The voltage gain of the
darlington transistors is less than unity in this case since they are
being used as emitter followers.

I am dropping 8.5 volts at 20 amps or 170 watts is being dissipated by
the transistors. This may not be a real problem. I know my commercial
power supply of similar size only provides about 18 volts to the
collector of the pass transistors and this �would result in about half
as much power being wasted. What voltage would you normally expect at
the the collector of the pass transistors. I have also thought that
maybe I dont have that much of a real problem since I doubt if the
power supply will ever be put under the same demands as it does
working into my load bank. Also I think I should add two more
transistors in paralell with the two I have. I think this would
distribute the heat better over my heat sink and have less localized
heating.

I dont think the size of my heatsink is a problem, it is probably
overkill. it is a 8in chimney type heat sink( i think that what you
call them) 8x8x10 inches finned inside and out. The transistor body is
mount on the outside �withe the wires going down the inside of the
chimney. The end of the chimney is designed to take a muffin fan.

Jimmie


I added two more pass transistors and this seems to have helped a lot.
I was once given a rule of thumb that a TO3 transistor should never
have to handle more than about 7 amps continously. I was pushing them
to 10amps. Even though they are rated at 50A they will not handle
anywhere near that without extreme measures to keep them cool.
Interesting note, at 40amps they dont get is hot with 4 transistors as
it did with 20amps with 2 transistors. I didnt check yet but I suspect
that is because the voltage at the ouput of the filter rectifier is
beginning to drop off a bit. If I continue to try to use this power
supply I would probably increase the number of pass transistors to 8.
This would be easy to do as the existing heatsink is already prepared
for it.

Thanks much for the advice
Jimmie