Searching \ for '[EE]: Power supply rework' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/power/actodc.htm?key=power
Search entire site for: 'Power supply rework'.

Exact match. Not showing close matches.
PICList Thread
'[EE]: Power supply rework'
2002\10\29@230030 by Dale Botkin

flavicon
face
part 1 1015 bytes content-type:TEXT/PLAIN; charset=US-ASCIIHi,

I just picked up a used radio with a power supply problem.  It's an
80s-vintage rig with a relatively simple but not too great power supply
section.  Specifically, it's a Kenwood TS-930S with a 307xxxx serial
number.  Where I should have 28.5V adjustable I get 33V non-adjustable,
which of course some parts of the radio are not too happy with.

I find all sort of documentation for fixing higher serial numbered rigs,
which have substantially different (but still not great) power supply
designs.  Nothing for one this old, though.  Rather than fix what is a
known to be a poor design to begin with, how does this look for a 28V
supply?  T1/T2 and R1/R2 are reused as is from the original.  In the
original, though, there are two transistors, a Zener, a pot and a few
other odds and ends driving the bases of T1 & T2.  I don't see a
compellign reason not to use a more modern voltage regulator, am I missing
something?

Dale
---
We are Dyslexia of Borg.
Fusistance is retile.
Your ass will be laminated.


part 2 5271 bytes content-type:IMAGE/png; name="pwrsupply.png" (decode)

part 3 154 bytes
--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2002\10\29@232201 by Rick C.

flavicon
face
Don't forget the 0.1uf bypass from reg input to adj term. Not using emitter
balancing resistors doesn't really share the load evenly although the base
resistors help a little. 0.33 ohm 10watt would suffice.
Rick

Dale Botkin wrote:

> am I missing
> something?
>
> Dale

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2002\10\29@233252 by Jim

flavicon
face
What are 317's rated for (Max V wise) these days?

Are there any conditions during startup (power-up)
of the supply where you have the possibility of
a damaging condition?

RF Jim

{Original Message removed}

2002\10\29@235753 by Dale Botkin

flavicon
face
On Tue, 29 Oct 2002, Rick C. wrote:

> Don't forget the 0.1uf bypass from reg input to adj term.

Naturally...  I didn't show them in the pic, but I always include them.

> Not using emitter balancing resistors doesn't really share the load
> evenly although the base resistors help a little. 0.33 ohm 10watt
> would suffice.

I was thinking of replacing the 2.2 Ohm base resistors with lower value
emitter resistors.  I may do that still, could even use the same mounting
points.

On Tue, 29 Oct 2002, Jim wrote:

> What are 317's rated for (Max V wise) these days?

+40/-3V input-output differential.

> Are there any conditions during startup (power-up) of the supply where
> you have the possibility of a damaging condition?

I'm trying to figure that out, and as you know I'm not a discrete
transistor kind of guy.  The transformer has two 28V secondaries,
parallelled and run to a full wave bridge for the 40VDC (high current).
The AC is also fed to a halfwave diode pair for the 33V supply (low
current).  According to the schematic (which I believe matches this SN
radio), the power switch turns on the AC into the transformer via relay.
There is a pair of 2200uF filter caps on the 40V side, so the 33V side
should always come up first, which would be a Good Thing.  On power off,
the caps will discharge thru the - well, I dunno, the 2N5885's I guess,
but what happens there?  I'm fuzzy on whether that could be seen at the
regulator output.  I'd think they could discharge thru the pot, but not if
the transistor base drive is gone...  Maybe a diode between the regulator
output and the transistor bases would be in order, then maybe a bleed
resistor across the caps?

The thing also has this all-discrete thermal fan control circuit.  I'm
thinking of replacing that with a DS1620 programmed for thermostatic
operation with hysteresis and use a small NPN or MOSFET to turn the fan on
and off.  That, or just hook the darned thing directly up to 8V and let it
run.

Thanks for the input!

Dale

--
http://www.piclist.com hint: PICList Posts must start with ONE topic:
[PIC]:,[SX]:,[AVR]: ->uP ONLY! [EE]:,[OT]: ->Other [BUY]:,[AD]: ->Ads


2002\10\30@003827 by Dale Botkin

flavicon
face
On Tue, 29 Oct 2002, Dale Botkin wrote:

> There is a pair of 2200uF filter caps on the 40V side, so the 33V side
> should always come up first, which would be a Good Thing.  On power off,
> the caps will discharge thru the - well, I dunno, the 2N5885's I guess,
> but what happens there?

Aha!  My bad.  There's a pair 2200uF caps on the 33V side too, which
should keep the regulator up and the transistors turned on long enough to
drain off the charge on the 40V filter caps.

Dale

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@030642 by Michael Rigby-Jones

flavicon
face
> -----Original Message-----
> From: Dale Botkin [SMTP:spam_OUTdaleTakeThisOuTspamBOTKIN.ORG]
> Sent: Wednesday, October 30, 2002 4:00 AM
> To:   .....PICLISTKILLspamspam@spam@MITVMA.MIT.EDU
> Subject:      [EE]: Power supply rework
> I don't see a compellign reason not to use a more modern voltage
> regulator, am I missing
> something?
>
> Dale
>
The only comment I would make is that your scheme throws away any current
limiting that the LM317 has.  What are the current requirements?  If the
LM317T is capable of supplying enough current (1.5 Amp ISTR?) then you could
put two in series with the first wired as a tracking pre-regulator.  I'm
almost certain the circuit is in the Nat-Semi datasheet.  Since LM317T's are
so cheap I'd always go for a scheme that has current limiting in if I could.

Regards

Mike

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@082305 by Dave Tweed

face
flavicon
face
Michael Rigby-Jones <mrjonesspamKILLspamNORTELNETWORKS.COM> wrote:
> > -----Original Message-----
> > From: Dale Botkin [SMTP:.....daleKILLspamspam.....BOTKIN.ORG]
> > Sent: Wednesday, October 30, 2002 4:00 AM
> > To:   EraseMEPICLISTspam_OUTspamTakeThisOuTMITVMA.MIT.EDU
> > Subject:      [EE]: Power supply rework
> > I don't see a compellign reason not to use a more modern voltage
> > regulator, am I missing
> > something?
> >
> > Dale
> >
> The only comment I would make is that your scheme throws away any current
> limiting that the LM317 has.

It throws away voltage regulation, too, so there's not much point in
continuing with this particular design.

An LM317 works by maintaining its own output 1.25 volts above the ADJ pin.
Normally, you take advantage of this by connecting a resistor of a few
hundred ohms between OUT and ADJ to create a constant current, which you
then pass through a control resistor to set the voltage of the ADJ pin,
which in turn sets the output voltage 1.25 V higher than that. The LM317
outputs a small (< 100 uA) and relatively constant current on the ADJ pin,
but this is normally less than 1% of the control current and can be
ignored.

With the proposed design, if the output voltage sags for any reason (e.g.,
increased load), the ADJ pin will be pulled down, and the base drive to the
NPN transistors will be reduced, not increased.

If you want to increase the current capacity of an LM317-based circuit, you
need to add a PNP transistor that is controlled by the current that the
LM317 draws into its IN pin. See the app notes. I'm not sure how you'd
incorporate the dual 33V/40V raw supplies -- maybe the PNP could then drive
your NPN pass transistors (with suitable emitter ballast resistors, of
course).

-- Dave Tweed

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@104643 by Jim

flavicon
face
  "It throws away voltage regulation, too"

I don't think his 'app' is going to care about
a percent or three 'sag' in output voltage ... he
isn't trying (nor does he need to) regulate
within .1% or anything close to it.

Design to spec and requirements - and no more.
To do so is to throw away both time and money ...

RF Jim

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@110512 by Dale Botkin

flavicon
face
On Wed, 30 Oct 2002, Jim wrote:

>    "It throws away voltage regulation, too"
>
> I don't think his 'app' is going to care about
> a percent or three 'sag' in output voltage ... he
> isn't trying (nor does he need to) regulate
> within .1% or anything close to it.

If that's all we're talking about, then no, it's really not an issue.

> Design to spec and requirements - and no more.
> To do so is to throw away both time and money ...

In this case, I'm just trying to at least match what the original (crappy,
late-70's Kenwood) design would do.  Precise regulation is really not an
issue, since the high-current spuuly just powers the final PA.  For that
matter, since I almost *never* run anywhere near the 125W or so output
this rig can produce, it's almost moot (I say almost because there are
occasions when I want to crank it up).  I could probably get by with a 5A
LM388S bolted to the heatsink and just stick to QRP, but I may want to
sell the rig some day.

Why would I use this massive piece of hardware for QRP (5W or less)
operation, you may ask?  Man...  you should hear this receiver.  It's
awesome.

Dale

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@113143 by Dave Tweed

face
flavicon
face
Dale Botkin <dalespamspam_OUTBOTKIN.ORG> wrote:
> On Wed, 30 Oct 2002, Jim wrote:
> >    "It throws away voltage regulation, too"
> >
> > I don't think his 'app' is going to care about
> > a percent or three 'sag' in output voltage ... he
> > isn't trying (nor does he need to) regulate
> > within .1% or anything close to it.
>
> If that's all we're talking about, then no, it's really not an issue.

I guess I was a little too subtle. The circuit will flat-out not work,
period. Have you ever breadboarded this configuration?

See page 13 of the following document for the proper circuit. You don't
have to use LM195s as the pass elements, but you will need to ballast the
emitters of ordinary NPN transistors appropriately, and perhaps adjust the
drive level of the PNP. I think that you should be able to feed the
collectors of the NPNs from a separate supply with no problem.

  http://www.national.com/ds/LM/LM117.pdf

-- Dave Tweed

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@115255 by Jim

flavicon
face
 "I guess I was a little too subtle. The circuit will
  flat-out not work,"


I would add a resistor between the output of the 317
and it's 3rd (reference) terminal - yielding a simple
'series' regulator (comprised of the two pass transistors)
using the 'usual' zener-to-ground circuit (effectively a
28 volt zener if the resistors are chosen correctly for
the 317).

The way he originally drew the circuit the Base-emitter
junctions are in series to R3 (his 10K adjustment pot).

His original circuit would still work - the Vbe drop
would have simply appeared in series with the drop
through R3 as the 317 tried to reach it's rated
output voltage between it's output and the 3rd
(Adj) terminal. (I can see wher at low output
currents the output voltage might have tended to
drift higher then the 28 V it was set for under load.)

RF Jim

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@123856 by Peter L. Peres

picon face
On Tue, 29 Oct 2002, Jim wrote:

*>What are 317's rated for (Max V wise) these days?

36V I think. There are schemes to run them higher (zener in series). The
datasheet contains application diagrams afair.

*>Are there any conditions during startup (power-up)
*>of the supply where you have the possibility of
*>a damaging condition?

You can't damage a 317 easily it has SOA and current limiting. Maybe a 337
would be better (TO3). I think that the unloaded input is 40V for 28V out
regulated.

Peter

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@125121 by Jim

flavicon
face
  "You can't damage a 317 easily "

The problem comes in when big caps (or when
caps charge from low-z sorces) supply loads
through unintentional paths (like when seen in
reverse bias conditions) that may pop the bond
wires inside the device package ... no SOA here.


(My discussion was concerned about start-up conditions,
not normal operating conditions where supervisory
funcitons inside the IC are allowed to perform their
proper functions!)

RF Jim

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@141551 by Peter L. Peres

picon face
On Wed, 30 Oct 2002, Jim wrote:

*>   "You can't damage a 317 easily "
*>
*>The problem comes in when big caps (or when
*>caps charge from low-z sorces) supply loads
*>through unintentional paths (like when seen in
*>reverse bias conditions) that may pop the bond
*>wires inside the device package ... no SOA here.

Maybe but a reverse connected diode across the device (output to input)
will do its job and in my experience 317s are extremely tough. The fact
that there is no high current path to ground helps. I had trouble with a
product (not mine) that supplied a troublesome connector from a 317T. The
connector would short in some units. I've never seen a failed 317 in any
of those, only the thermal fuse in the transformer went.  These were 800mA
supplies I think. The internal protection in the 317 is pretty good even
if you put a dead short across the output. It's SOA + overcurrent
protection (probably by controlled beta or something like that). The only
way to kill it is overvoltage ('back' voltage) from the output to the
input. SLA batteries can do this if the bridge or other parts upstream are
shorted (in a charger application). Otherwise it is very hard to kill
them.

Peter

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@145142 by Jim

flavicon
face
  "Otherwise it is very hard to kill them."

Again, I hope you didn't miss the point. Most devices
do work *quite* well when operated within their design
specs (who would argue with that?). But THAT is just the
beginning ... SOMETIMES they treated improperly (like
during power up and power down conditions, operation over
temperature, TRANSIENT response, RF susceptability, etc).

I would like to state future reference that I have been
'round the bend with simple DC power supplies (since before
the 317 was invented even!). Back nearly two decades ago now
I even played a key role in solving a power supply 'dynamics'
problem for which I received a letter of commendation from
Dave Nurse (President at the time of Heathkit in St. Joseph -
Benton Harbor was just the mailing address). This stands as
testament to the solution of a nagging problem that had
plauged their SB-104 power supply for years and for which
*no* adequate solution had been found - until I encountered
the problem.

Ever heard of "charge storage" (excess carriers having been
injected into the base) effectrively occuring in the BE junction
of a bipolar transistor - a condition that can lead to excessive
turn-off delay times *after* a device has been driven far into
saturation?

All those who FINALLY got instructions on what mods to make
to get their Heath SB-104 power supplies to *stop* blowing
fuses (because the OVP circuit tripped) when coming out
of the 'Tune' position (or simply powering on the radio)
owe me thanks for finding 'the cure' to this nagging problem.

RF Jim (specializing not only in RF but DC as well)

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@152129 by Dave Tweed

face
flavicon
face
Jim <@spam@jvpollKILLspamspamDALLAS.NET> wrote:
>   "I guess I was a little too subtle. The circuit will
>    flat-out not work,"
>
> The way he originally drew the circuit the Base-emitter junctions are in
> series to R3 (his 10K adjustment pot). His original circuit would still
> work - the Vbe drop would have simply appeared in series with the drop
> through R3 as the 317 tried to reach it's rated output voltage between
> it's output and the 3rd (Adj) terminal.

But you're ignoring the current gain of the transistor, which introduces a
huge amount of positive feedback, completely swamping the negative feedback
that normally regulates the output voltage.

The LM317 is designed to operate with zero impedance between its OUT
terminal and the load. If you call the top half of the adjustment pot R1
and the bottom half R2 (following National's notation), any resistance
between OUT and load is effectively multiplied by the factor 1 + R2/R1.

Also, the voltage across R1 is now only 0.5 to 0.6 V, so the R2:R1 ratio is
about 50:1. Even if the effective resistance of the BE junctions and the
ballast resistors is on the order of 1 ohm, the power supply will exhibit a
source resistance on the order of 50 ohms, divided by the current gain of
the pass transistors. If the latter is on the order of 20-50, we're left
with an overall value of 1-2 ohms.

With a load that can go up to 10A, this gives a voltage drop of 10-20V (out
of 28). I think we can safely say that this circuit is NOT regulating the
voltage.

The above discussion assumes that the Vbe of the transistors is more or
less constant, but I see from the data sheet that it can rise as high as
1.5V (at a collector current of 10A), completely swamping the output of the
LM317.

Note also that an LM317 by itself exhibits a source resistance on the order
of 0.01 ohm.

> I would add a resistor between the output of the 317 and it's 3rd
> (reference) terminal - yielding a simple 'series' regulator (comprised of
> the two pass transistors) using the 'usual' zener-to-ground circuit
> (effectively a 28 volt zener if the resistors are chosen correctly for
> the 317).

Yes, if you also break the feedback path from the NPN emitters. Now you
simply have a regulated voltage source with an emitter follower for current
gain. The output voltage variation from no load to full load should be less
than about 2V, almost all of it due to the ballast resistors (assuming 0.33
ohms for each emitter).

I still say it would be better to use the circuit given in the app note.

> (I can see wher at low output currents the output voltage might have
> tended to drift higher then the 28 V it was set for under load.)

Yes, the minimum load on the power supply would have been the quiescent
current of the LM317 (about 5 mA worst case, normally absorbed by the
feedback network) multiplied by the current gain of the transistors at low
collector currents (100 or more), or about 500 mA.

-- Dave Tweed

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@155229 by Dale Botkin

flavicon
face
part 1 600 bytes content-type:TEXT/PLAIN; charset=US-ASCIIOn Wed, 30 Oct 2002, Peter L. Peres wrote:

> Maybe but a reverse connected diode across the device (output to input)

Planned.  National has a specific recommendation on this for all of their
3-terminal regulators, and I was planning to add a few 1N4001s, since I
have bags of 'em.  See attached.

Jim, is this what you were talking about (note R4 and the protection
diodes D1/D2) and Dave, do you think this will work?  I can see National's
data sheet solution for this, but I'm not clear on why they're showing it
that way.  I still am not showing the caps but they are always there.

Dale





part 2 2250 bytes content-type:IMAGE/png; name="TS930pwr.png" (decode)

part 3 144 bytes
--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@164320 by Dave Tweed

face
flavicon
face
Dale Botkin <KILLspamdaleKILLspamspamBOTKIN.ORG> wrote:
> Jim, is this what you were talking about (note R4 and the protection
> diodes D1/D2) and Dave, do you think this will work?

Yes, it should work, but I would still disconnect the top end of the pot
from the output terminal and just use it as a variable resistor.

With the 240 ohm resistor in there, the pot will be set to about 5600 ohms
to ground on the bottom half. This puts the ADJ pin at 28V, the OUT pin at
29.25V and the output at around 28.5V. The top half of the pot will only
have about half a volt acrosss it, but this still represents a source of
counter-productive positive feedback, especially for dynamic load changes.

Also, I think that 0.01 ohm ballast resistors are about an order of
magnitude too low to properly balance the current between the two pass
transistors. At 5A max per leg, they'll only drop about 50 mV, which isn't
much feedback for the transistors. On the other hand, either transistor
alone can handle the peak current, so maybe it doesn't matter that much.

> I can see National's data sheet solution for this, but I'm not clear on
> why they're showing it that way.

Rather than explaining it again, can we just assume they know what they're
doing? Wouldn't you rather have 10x better regulation? One more transistor
and a couple of resistors is a small price to pay for this.

Basically, the PNP senses the current that the LM317 is using to control
the load, and diverts most of it through the external pass transistors
instead. Since the sense circuit is on the input side of the regulator,
it doesn't affect the basic voltage feedback mechanism.

-- Dave Tweed

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@222230 by Jim

flavicon
face
part 1 5844 bytes content-type:text/plain; (decoded 7bit)

Comments within:

----- Original Message -----
From: "Dave Tweed" <RemoveMEpicTakeThisOuTspamDTWEED.COM>
To: <spamBeGonePICLISTspamBeGonespamMITVMA.MIT.EDU>
Sent: Wednesday, October 30, 2002 2:18 PM
Subject: Re: [EE]: Power supply rework


{Quote hidden}

feedback

Fortunately - we're dealing with transistors in the
"Emitter Follower configuration" here (whose voltage
gain is less than one) and therefore rather harmless
(the dangers being when they happen to upset DC bias
conditions that we need or want at some point in order
to make the circuit work) ...

> that normally regulates the output voltage.
>
> The LM317 is designed to operate with zero impedance between its OUT
> terminal and the load.

1) Where do you get that?

2) The application of ANY kind of load suddenly presents a
  complex load NOT comprised of any given or assumed
  impedance ...

If you are thinking that the "output pin" must be connected
to the load in order for the 317 to regulate at it's rated
spec - you would be right (see the attached functional block
diagram for reference) ... BUT that does not preclude anyone
from using it as a 'power zerer' to drive an NPN transistor
(or two) for a simple series-regulator power supply ...


> terminal and the load. If you call the top half of the adjustment pot R1
> and the bottom half R2 (following National's notation), any resistance
> between OUT and load is effectively multiplied by the factor 1 + R2/R1.
>
> Also, the voltage across R1 is now only 0.5 to 0.6 V, so the R2:R1 ratio
is
> about 50:1. Even if the effective resistance of the BE junctions and the
> ballast resistors is on the order of 1 ohm, the power supply will exhibit
a
> source resistance on the order of 50 ohms, divided by the current gain of
> the pass transistors.

Now you're beginning to see the light ...

> the pass transistors. If the latter is on the order of 20-50, we're left
> with an overall value of 1-2 ohms.
>
> With a load that can go up to 10A, this gives a voltage drop of 10-20V
(out
> of 28). I think we can safely say that this circuit is NOT regulating the
> voltage.

I think the load current is going to be in 10A range max (that
would be on the order of 280 watts of DC input power to Dale's
rig when operated at full tilt for 120W out - somewhat close
to what these HF radios seem to require),

>
> The above discussion assumes that the Vbe of the transistors is more or
> less constant, but I see from the data sheet that it can rise as high as
> 1.5V (at a collector current of 10A), completely swamping the output of
the
> LM317.
>

I didn't perform a full DC analysis - Dale didn't supply an 'EO'
for that! (An EO is what TI used to call an 'engineering order'
and represented a 'charge number' against which time and
materials could be billed.)

> Note also that an LM317 by itself exhibits a source resistance on the
order
> of 0.01 ohm.
>
> > I would add a resistor between the output of the 317 and it's 3rd
> > (reference) terminal - yielding a simple 'series' regulator (comprised
of
> > the two pass transistors) using the 'usual' zener-to-ground circuit
> > (effectively a 28 volt zener if the resistors are chosen correctly for
> > the 317).
>
> Yes, if you also break the feedback path from the NPN emitters. Now you
> simply have a regulated voltage source with an emitter follower for
current
> gain. The output voltage variation from no load to full load should be
less
> than about 2V, almost all of it due to the ballast resistors (assuming
0.33
> ohms for each emitter).
>
> I still say it would be better to use the circuit given in the app note.

Probably ... there are several other designs I would go
with (that would also incorporate some fold-back current
limiting as well!).

I was looking at circuit in the data sheet, the one using
the LM195's in parallel (that you brought up) and also the
one which shows adding an LM301A that incorporates an
adjustable current limit (on page 14 of the app note).

(That -8V negative voltage could be ignored and instead
just tie that pin of the LM301A to ground since Dale
is goint to set this thing up for 28 some volts out.)

For reference:

AN-182: Application Note 182 Improving Power Supply Reliability with IC
Power Regulators
http://www1.national.com/an/AN/AN-182.pdf

Applications for an Adjustable IC Power Regulator (AN-178):
www.aet.cup.edu/ftp/eet/datasheets/National_Semiconductor_Application
_Notes/an-178.pdf

117 regulator data sheet:
http://www.national.com/ds/LM/LM117.pdf

DC Jim

{Quote hidden}


part 2 9686 bytes content-type:image/gif; (decode)


part 3 144 bytes
--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\30@223024 by Jim

flavicon
face
part 1 1067 bytes content-type:text/plain; (decoded 7bit)

Better Dale ... the regulation should be better
with this ...

How adverse are you to complexity, Dale (kindly see
the attached shematic)?

DC Jim

{Original Message removed}
part 2 12946 bytes content-type:image/gif; (decode)


part 3 144 bytes
--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics


2002\10\31@025735 by Michael Rigby-Jones

flavicon
face
> -----Original Message-----
> From: Dale Botkin [SMTP:RemoveMEdalespamTakeThisOuTBOTKIN.ORG]
> Sent: Wednesday, October 30, 2002 8:52 PM
> To:   PICLISTEraseMEspam.....MITVMA.MIT.EDU
> Subject:      Re: [EE]: Power supply rework
>
>
> Jim, is this what you were talking about (note R4 and the protection
> diodes D1/D2) and Dave, do you think this will work?  I can see National's
> data sheet solution for this, but I'm not clear on why they're showing it
> that way.  I still am not showing the caps but they are always there.
>
> Dale
>
>
I don't understand why your pass transistors are fed from a 40 volt supply
and the LM317T from a 33 volt supply?  Does the 40v supply have a higher
current capability that is required?

If you need high current then the method shown in the back of the datasheet
using a sense resistor is superior.  You could combine this with the HV
version of the LM317 to get a minimal parts count solution.

Regards

Mike

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\31@025943 by Michael Rigby-Jones

flavicon
face
> -----Original Message-----
> From: Jim [SMTP:EraseMEjvpollspamDALLAS.NET]
> Sent: Wednesday, October 30, 2002 3:47 PM
> To:   RemoveMEPICLISTEraseMEspamEraseMEMITVMA.MIT.EDU
> Subject:      Re: [EE]: Power supply rework
>
>
> Design to spec and requirements - and no more.
> To do so is to throw away both time and money ...
>
> RF Jim
>
Although I agree in the general case, if a superior design can be realised
with the same or less components, then why not do it?

Mike

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\31@030428 by Michael Rigby-Jones

flavicon
face
> -----Original Message-----
> From: Jim [SMTP:RemoveMEjvpollspam_OUTspamKILLspamDALLAS.NET]
> Sent: Wednesday, October 30, 2002 7:51 PM
> To:   RemoveMEPICLISTTakeThisOuTspamspamMITVMA.MIT.EDU
> Subject:      Re: [EE]: Power supply rework
>
> All those who FINALLY got instructions on what mods to make
> to get their Heath SB-104 power supplies to *stop* blowing
> fuses (because the OVP circuit tripped) when coming out
> of the 'Tune' position (or simply powering on the radio)
> owe me thanks for finding 'the cure' to this nagging problem.
>
> RF Jim (specializing not only in RF but DC as well)
>
Jeez, I sure hope they have wide doorways at Dallas. :o)

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\31@041713 by john

flavicon
face
the one diode is to pass the residual output voltage back to the input ( that
way you should never get more than ~0.7 v output to input differential if the
outputs storage capacity is greater than the input ... load spikes ... blah
blah blah) so that you dont reverse voltage the main pass tranny inside the
317.
the other is to help discharge the cap on the adjust terminal safely, instead
of through the regulators adj pin.
i see yours has no cap ? its wise to put a nice tant on there, helps with
noise and spikes

On Wednesday 30 October 2002 10:52 pm, you wrote:
{Quote hidden}

--
Thank-you for your time.

John Ward

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\31@063447 by Alan B. Pearce

face picon face
>With a load that can go up to 10A, this gives a voltage
>drop of 10-20V (out of 28). I think we can safely say
>that this circuit is NOT regulating the voltage.

Well, to be pedantic, it is, in that it is limiting the maximum voltage
under no load, so it is performing as well as a shunt zener regulator :))

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\31@111334 by Dale Botkin

flavicon
face
On Wed, 30 Oct 2002, Dave Tweed wrote:

8< snipped some pretty good stuff...

OK, of course I see this 30 seconds after sending my last email, which
will now look pretty bad I'm sure.

In reding the data sheet for the LM317 and the LM388 the first time
through, I was thinking in terms of a 7805-type regulator, where the
output is maintained at X number of Volts above the reference pin.  I can
see that's not the case.

> With a load that can go up to 10A, this gives a voltage drop of 10-20V (out
> of 28). I think we can safely say that this circuit is NOT regulating the
> voltage.

It'll be over 10A per transistor, the total load may be well over 20A
under full power I think.

> Yes, if you also break the feedback path from the NPN emitters. Now you
> simply have a regulated voltage source with an emitter follower for current
> gain. The output voltage variation from no load to full load should be less
> than about 2V, almost all of it due to the ballast resistors (assuming 0.33
> ohms for each emitter).

That works for me.  In fact I *think*, near as my poor dumb non-analog,
non-EE head can figure it out, that's what was originally there.  I could
be wrong.

I do know I spent several hours last night dissecting the regulator
circuit board and testing components.  There's a close approximation at
http://www.jvgavila.com/ts930_ps.gif, though it's of a later version.  In mine,
the emitter of Q1 is fed by D1 & D2 instead of the 40V supply, and the
whole Q3/Q6 part is not there.  I didn't find anything defective (didn't
test them ALL, but all the semis).  Resoldered all the joints, and in the
end nothing changed.  Due to very tight clearances I can't get probes in
to look at things while it's powered on, except for the base of Q2.  That
changes when I adjust the pot VR1 (15 to 18V), but nothing else does.
The junction of D3 and D6 stays at about 33V, as does BS1 & BS2 on the #3
connector which goes to the two 2N5885 power transistors.  Seems like
everywhere I look I see 33V plus or minus a little.  I think I'm going to
probably just rip out all but the power transistors and use an LM317 to
drive the transistor bases and see what happens.  I hate not knowing
what's wrong with it though.

By the way, the current design just uses R4 & R5 to the bases of the two
big transistors, the emitters & collectors are paralleled.  I'll probably
do the same, since that part seems to work.

Dale

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\31@165943 by Jim

flavicon
face
Credit where credit is due, I say.

No more and certainly *no less*!

RF Jim

----- Original Message -----
From: "Michael Rigby-Jones" <EraseMEmrjonesspamspamspamBeGoneNORTELNETWORKS.COM>
To: <RemoveMEPICLISTKILLspamspamMITVMA.MIT.EDU>
Sent: Thursday, October 31, 2002 2:02 AM
Subject: Re: [EE]: Power supply rework


> > {Original Message removed}

2002\10\31@193929 by Dave Tweed

face
flavicon
face
Jim <jvpollSTOPspamspamspam_OUTDALLAS.NET> wrote:
> > The LM317 is designed to operate with zero impedance between its OUT
> > terminal and the load.
>
> 1) Where do you get that?

From the data sheet.

> Now you're beginning to see the light ...

I've been seeing the light all along. It was you who needed convincing.
Since you didn't contradict anything in my post, I assume you now basically
agree with my points.

> I didn't perform a full DC analysis - Dale didn't supply an 'EO' for that!

Then why did you go to so much trouble to express support for a design that
is counter to National's recommendations?

Anyway, I'm finished with this topic. No hard feelings?

-- Dave Tweed

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\31@211410 by Russell McMahon

face
flavicon
face
> In reading the data sheet for the LM317 and the LM388 the first time
> through, I was thinking in terms of a 7805-type regulator, where the
> output is maintained at X number of Volts above the reference pin.  I can
> see that's not the case.

This is more an aside in this instance, but worth noting for future
reference (no pun intended :-) ).
The LM317 and the 7805 essentially DO work identically - it's just that they
are usually *used* differently and are each optimised for their typical and
usually slightly different applications.
The 7805 seeks to maintain 5v between the output and reference pin. The load
is usually connected across the same points so it is true to say, as you do,
" ... the output is maintained at X number of Volts above the reference
pin."

Now the LM317 is a 1.2 volt regulator. It seeks to maintain the output pin
1.2v above the reference pin. However, 1.2v is not usually an especially
useful voltage so the regulator is used in configurations such that WHEN the
LM317 maintains this desired difference, the consequent voltage produced
across the load is equal to some other voltage - say 5v or 12v or whatever.
This is (usually) achieved by dividing the actual output resistively such
that a fraction of it appears across the output pins of the LM317. When Vout
is the desired voltage the LM317 sees its target voltage and all is well.
When the target output voltage drops or rises the LM317 sees a reduction or
increase in its 1.2v output level and adapts accordingly. So it is also true
to say for the LM317 that " ... the output is maintained at X number of
Volts above the reference pin." It's just that in this case, the reference
pin is NOT at earth, so the load sees this X volts plus the voltage from
reference pin to ground.

The LM317 designers did not, of course, set out to make a regulator for use
at 1.2v. They wanted to make a regulator that could be used for a wide range
of voltages by using division of the output as described above. The 1.2v
figure is simply a figure that is as low as possible while still providing
enough internal drop to allow the internal electronics to function. (It
needs to be low as the regulator can then be used for voltages equal to or
more than this voltage) The 7805 and any other similar 3 terminal regulator
CAN be used in a similar manner to produce a voltage higher than its design
voltage by dividing the load voltage so the target voltage is eg 5v when the
design voltage is at target voltage. (eg take a 7805 and place 1k from
output to its "ground" pin and 1k from this point to true ground and you
will get "about" 10 volts out.) Such regulators are usually less successful
than purpose designed ones such as the LM317 because they have not been
designed to take account of the effect of the resistor from reference to
ground. On eg 7805 the internal electronics use this path for their main
supply current and this affects the regulation as this current varies. In
the eg LM317 the internal electronics are largely powered from the in/out
voltage drop and the ref current is kept constant. This greatly improves
regulation.

None of which is too important in this case, but it's worth being aware of
for future reference.


       RM

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.


2002\10\31@225254 by Dave Tweed

face
flavicon
face
Dale Botkin <spamBeGonedaleSTOPspamspamEraseMEBOTKIN.ORG> wrote:
> On Wed, 30 Oct 2002, Dave Tweed wrote:
> > With a load that can go up to 10A, this gives a voltage drop of 10-20V (out
> > of 28). I think we can safely say that this circuit is NOT regulating the
> > voltage.
>
> It'll be over 10A per transistor, the total load may be well over 20A
> under full power I think.

Wow. I was assuming that the final would be something just under 50%
efficient, so I was guessing that 280W input power would be more than
enough for 125W out.

{Quote hidden}

Odd circuit. If all the semiconductors tested OK, there must be an open
ground somewhere, maybe R6 (the 1.2K at the emitter of Q2).

However, notice how much the topology of the circuit resembles the
high-current adjustable regulator based on the LM317.

Take out Q2 and change R7 (at the top end of VR1) to about 100 ohms.

Connect the LM317 as follows:
 IN to the base of Q1
 OUT to the anode of D8 (the output bus)
 ADJ to the wiper of VR1

You might need to fiddle with the value of R2 in order to give the LM317
enough current to work with. Try changing it to 4700 ohms.

You can also take out D3, D6, D8 and R6 since they aren't used, but it
doesn't hurt anything to leave them in.

> By the way, the current design just uses R4 & R5 to the bases of the two
> big transistors, the emitters & collectors are paralleled.  I'll probably
> do the same, since that part seems to work.

Whatever. If it worked before, it should continue to work.

-- Dave Tweed

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.



'[EE]: Power supply rework'
2002\11\01@004406 by Dale Botkin
flavicon
face
On Thu, 31 Oct 2002, Dave Tweed wrote:

> > It'll be over 10A per transistor, the total load may be well over 20A
> > under full power I think.
>
> Wow. I was assuming that the final would be something just under 50%
> efficient, so I was guessing that 280W input power would be more than
> enough for 125W out.

Duh, never mind.  28V, not 13.8...  yeah, 10A is right.  Duh.  Sorry about
that.

Dale

--
http://www.piclist.com#nomail Going offline? Don't AutoReply us!
email KILLspamlistservspamBeGonespammitvma.mit.edu with SET PICList DIGEST in the body


More... (looser matching)
- Last day of these posts
- In 2002 , 2003 only
- Today
- New search...