Maker Pro
Maker Pro

Redesign, was low voltage drop transistors for power supply

A

Albert

Jan 1, 1970
0
OK, thanks to all who provided some guidance in the original post (see
'low voltage drop transistors for power supply' post.

Clearly, this was a 12v supply and not a 13.8v supply. The actual
output current ratings are not known, the only hint is that the
original meter was a 25 amp meter.

The transformer is much larger than 100 va and is actually larger than
a 200 va unit (based on core size). When I tore down the rest of the
assembly, I found more shaky stuff including sockets for the
transistors that were soldered, several etch cuts and lifted run on
the PCB, meter shunt was missing (probably discarded when the original
pcb was replaced). In other words, it probably needs a complete
redesign rather than using any of the original parts and pc boards.

I ran LTC's power supply design software and it came up with a 13.8
volt output switching supply that will supply 14A with the existing
power transformer. It uses the LTC1775 switching regulator, 2 X
Si4410DY mosfets, an 11 uH inductor and a couple of 1N5818 schottky
diodes. The supply itself (not including transformer losses and
rectifier losses) provides 98 percent efficiency at 14A output.

So, I'm considering building up this switching supply.

I need suggestions for a schottkey bridge rectifier that will handle
20A or more. I find the diodes themselves are readily available, but
I'd like to have a bridge assembly if possible.

Any suggestions??
 
L

legg

Jan 1, 1970
0
I need suggestions for a schottkey bridge rectifier that will handle
20A or more. I find the diodes themselves are readily available, but
I'd like to have a bridge assembly if possible.

The cheapest way to increase the current handling capacity of a
conventional bridge is to use only one half of two seprate bridges.

If it's the forward voltage drop only that you are worried about, the
industry habit is to drop a full-wave bridge for a center-tap
topology.

Schottkys with fast-on or screw terminals are expensive unless you can
find an older module on the surplus market. Semtech used to do some
schottky versions that I don't see listed http://www.semtech.com/ .
The most likely recent package offered is a variant of TO-227, which
harbours two diodes per package with four terminals. This can be
configured by the end user as CC, CA or doubler.

ST Micro
http://www.st.com/stonline/bin/sftab.exe?type=&table=541&filter-XJE010_def=STPS16045

Microsemi
http://www.microsemi.com/catalog/pack.asp?packid=SOT-227

IXYS
http://www.ixys.com/Appasp/pddiod03p.asp

Though IR announced that they intended to use this package in '97,
they've never followed up on it seriously and it has never become a
commodity package - remaining in the region of $20 a pop.

There are many cheaper through-hole pcb-mounting discrete diode
alternatives for 20A that you should probably consider first.

If you are adamant in using the original transformer to produce 13V5,
you might at least choose a switching regulator topology capable of
doing the job with the voltage available. If you connected your 60Hz
supply as a negative voltage, you could regulated a positive 13v5
using a simple boost topology without special consideration for the
input voltage (actually an inversion of the buck-boost topology).

Switchers make noise. If this is for RF you should reinspect the box
and I/O connections for EMI suppression capability.

RL
 
R

R.Legg

Jan 1, 1970
0
Albert said:
I ran LTC's power supply design software and it came up with a 13.8
volt output switching supply that will supply 14A with the existing
power transformer. It uses the LTC1775 switching regulator, 2 X
Si4410DY mosfets, an 11 uH inductor and a couple of 1N5818 schottky
diodes. The supply itself (not including transformer losses and
rectifier losses) provides 98 percent efficiency at 14A output.

So, I'm considering building up this switching supply.

Rerun the calculations using the output of your 60Hz transformer as a
full-wave doubler. (uses only two diodes, but needs two capacitors).
This should produce a no-load voltage less than the LTC1775's input
rating of 36V, based on your previous measurements for full-wave no
load.

I think you'll find that raising the input voltage to the converter
increases the potential continuous output rating, as the duty of the
mosfets is closer to 50% - unless this was the product of the peak
current limiting function of the controller alone.

If your load demands higher current, it might not respond the same as
it used to with the linear, which simply went out of regulation while
producing the needed juice.

RL
 
N

N. Thornton

Jan 1, 1970
0
Albert said:
Clearly, this was a 12v supply and not a 13.8v supply. The actual
output current ratings are not known, the only hint is that the
original meter was a 25 amp meter.

The transformer is much larger than 100 va and is actually larger than
a 200 va unit (based on core size). When I tore down the rest of the
assembly, I found more shaky stuff including sockets for the
transistors that were soldered, several etch cuts and lifted run on
the PCB, meter shunt was missing (probably discarded when the original
pcb was replaced). In other words, it probably needs a complete
redesign rather than using any of the original parts and pc boards.

I ran LTC's power supply design software and it came up with a 13.8
volt output switching supply that will supply 14A with the existing
power transformer. It uses the LTC1775 switching regulator, 2 X
Si4410DY mosfets, an 11 uH inductor and a couple of 1N5818 schottky
diodes. The supply itself (not including transformer losses and
rectifier losses) provides 98 percent efficiency at 14A output.

So, I'm considering building up this switching supply.

I need suggestions for a schottkey bridge rectifier that will handle
20A or more. I find the diodes themselves are readily available, but
I'd like to have a bridge assembly if possible.

Any suggestions??


Of course I've not seen the unit but it strikes me that sorting those
bugs out may be quicker than a complete rebuild - they dont sound like
hard to fix issues.

The main plus with that is you get out the full amount of oomph that
your power TF is capable of giving. If its got a 25A meter on it,
chances are it can give at least 20A, and probably 25A. Other plus is
its less work.


Regards, NT
 
Albert said:
OK, thanks to all who provided some guidance in the original post (see
'low voltage drop transistors for power supply' post.

Clearly, this was a 12v supply and not a 13.8v supply. The actual
output current ratings are not known, the only hint is that the
original meter was a 25 amp meter.

The transformer is much larger than 100 va and is actually larger than
a 200 va unit (based on core size). When I tore down the rest of the
assembly, I found more shaky stuff including sockets for the
transistors that were soldered, several etch cuts and lifted run on
the PCB, meter shunt was missing (probably discarded when the original
pcb was replaced). In other words, it probably needs a complete
redesign rather than using any of the original parts and pc boards.

I ran LTC's power supply design software and it came up with a 13.8
volt output switching supply that will supply 14A with the existing
power transformer. It uses the LTC1775 switching regulator, 2 X
Si4410DY mosfets, an 11 uH inductor and a couple of 1N5818 schottky
diodes. The supply itself (not including transformer losses and
rectifier losses) provides 98 percent efficiency at 14A output.

So, I'm considering building up this switching supply.

I need suggestions for a schottkey bridge rectifier that will handle
20A or more. I find the diodes themselves are readily available, but
I'd like to have a bridge assembly if possible.

Any suggestions??
Yup. Abandon the switcher idea if this is to be used
for a ham station. Put two decent pass transistors in
there. If you are dead set on a redesign, use an LM317
with the pass transistors in collector wrap around
configuration. If you don't want it adjustable, use
a 7812 as if it was an LM317 (gnd leg is the adj leg),
with the resistors chosen to give you 13.8. If it can't
sustain 13.8 across the supply output at full load, use
Watson's suggested boost circuit. When you have that
working to your satisfaction, add a crowbar.

The down-side of the above is not getting to play
with the switcher design. The down-side of the switcher
is noise.
 
R

R.Legg

Jan 1, 1970
0
Of course I've not seen the unit but it strikes me that sorting those
bugs out may be quicker than a complete rebuild - they dont sound like
hard to fix issues.

The main plus with that is you get out the full amount of oomph that
your power TF is capable of giving. If its got a 25A meter on it,
chances are it can give at least 20A, and probably 25A. Other plus is
its less work.

This is called 'shooting the engineer', and sometimes it's a good
idea. I assume the original owner really just wants it to do what it
did, before the fuse went. If you can fit in some small improvements
(like reducing 'chirp'), that don't cause a lot of delay, then it
might be best just to get it off of your bench asap.

I shouldn't give the meter FSD much credence, though.

RL
 
P

Peter A Forbes

Jan 1, 1970
0
This is called 'shooting the engineer', and sometimes it's a good
idea. I assume the original owner really just wants it to do what it
did, before the fuse went. If you can fit in some small improvements
(like reducing 'chirp'), that don't cause a lot of delay, then it
might be best just to get it off of your bench asap.

I shouldn't give the meter FSD much credence, though.

RL

We've seen moving iron 'output amps' meters in the ac circuit to the bridge
rectumfrier, thus giving a rather inflated view of the output current....

Don't forget that the available DC output current from a full wave bridge with
smoothing cap is only about 0.6 of the AC current, although the DC Volts is 1.44
X (in theory)

Peter
 
P

Peter A Forbes

Jan 1, 1970
0
If you don't want it adjustable, use
a 7812 as if it was an LM317 (gnd leg is the adj leg)

That's a bit ambiguous, the pinouts in the TO220 variants go thus:

78XX = output / gnd / input
LM317 = input / output / adj

The tab on the 78XX is Gnd while on the LM317T it is the output.

Data sheets are on the web from NatSemi and others for verification.

Peter
 
N

N. Thornton

Jan 1, 1970
0
This is called 'shooting the engineer', and sometimes it's a good
idea. I assume the original owner really just wants it to do what it
did, before the fuse went. If you can fit in some small improvements
(like reducing 'chirp'), that don't cause a lot of delay, then it
might be best just to get it off of your bench asap.


shooting the engineer?
reducing chirp on a PSU?
PSU delay?
Just call me puzzled :)

Regards, NT
 
R

R.Legg

Jan 1, 1970
0
Peter A Forbes said:
We've seen moving iron 'output amps' meters in the ac circuit to the bridge
rectumfrier, thus giving a rather inflated view of the output current....

Don't forget that the available DC output current from a full wave bridge with
smoothing cap is only about 0.6 of the AC current, although the DC Volts is 1.44
X (in theory)

I thought about this as a suggestion but couldn't remember the
relationship for average current in this position. Peak to average
ratio is quite high here so anything measuring rms wouldn't give a
good output DC current indication.

Moving iron meters are pretty rare.

Moving coil meters indicate average current, so the error wouldn't be
large.

RL
 
R

R.Legg

Jan 1, 1970
0
Peter A Forbes said:
We've seen moving iron 'output amps' meters in the ac circuit to the bridge
rectumfrier, thus giving a rather inflated view of the output current....

Don't forget that the available DC output current from a full wave bridge with
smoothing cap is only about 0.6 of the AC current, although the DC Volts is 1.44
X (in theory)

I meant the error in an average-responding meter wouldn't be large
after the rectifier.

RL
 
J

Jan Panteltje

Jan 1, 1970
0
terminating the design phase - producing the goods.
Yea that makes sense, I remember 25 years ago we had 500 units of something
I designed ready to ship, and I was re-calculating things on a piece of paper,
and told the boss 'I think we need to make this resistor 1/2 Watt, now it is
1/4, it could perhaps overheat ... formula... heatwave... solar blasts...
nukes...'
He just looked at me, I got the message.
JP
 
T

Tim Wescott

Jan 1, 1970
0
Yup. Abandon the switcher idea if this is to be used
for a ham station. Put two decent pass transistors in
there. If you are dead set on a redesign, use an LM317
with the pass transistors in collector wrap around
configuration. If you don't want it adjustable, use
a 7812 as if it was an LM317 (gnd leg is the adj leg),
with the resistors chosen to give you 13.8. If it can't
sustain 13.8 across the supply output at full load, use
Watson's suggested boost circuit. When you have that
working to your satisfaction, add a crowbar.

The down-side of the above is not getting to play
with the switcher design. The down-side of the switcher
is noise.

His original problem was a very low overhead voltage out of the
rectifier -- using a high drop-out regulator would be a _bad_ idea.
 
K

KR Williams

Jan 1, 1970
0
Yea that makes sense, I remember 25 years ago we had 500 units of something
I designed ready to ship, and I was re-calculating things on a piece of paper,
and told the boss 'I think we need to make this resistor 1/2 Watt, now it is
1/4, it could perhaps overheat ... formula... heatwave... solar blasts...
nukes...'
He just looked at me, I got the message.

Yeah, if you came in with that news that late, I'd be looking at
you with an evil eye too. Our rules were always at *least* a 60%
under-rating. One was wary of ccoming even close to that because
the chance of failure (and reprisal) was simply too great. The
requirement was 100K POH (and of course a 2% resistor was treated
as 10%, etc.). End-of-life and worst-case were drilled into the
design culture. Not so much anymore...
 
Tim said:
His original problem was a very low overhead voltage out of the
rectifier -- using a high drop-out regulator would be a _bad_ idea.

No, the original problem per the poster was a burned out
supply. The poster's description included the impossibility
of a PNP in parallel with an NPN as the pass transistors,
so someone was in there with a "golden screwdriver" before
he got it, or his post was incorrect.

Regarding headroom/low drop out:
The LM317 needs a bit less than 2 volts headroom at
1 amp, per the datasheet. With the "Watson boost circuit"
(if needed), he'll get about 1.9 volts extra or about
16.9 total, assuming his 15 volt measurement was accurate.
That's 3.1 volts headroom. The 7812 datasheet shows a
slightly lower headroom requirement.
 
T

Tam/WB2TT

Jan 1, 1970
0
Albert,

I have looked at all the responses to this, and your other posting. Probably
the simplest solution is to use the brute force solution of adding a 6V 1A
transformer, as somebody suggested.

Put a bridge rectifier and filter on this transformer to give you about
8VDC.

Connect the negative side of this supply to your 17V.

The positive side of this rectifier will now be your boosted B+ of about
25V.

Run your output transistors off the 17V. Change them to 2N3771 or better.
While you are at it, you may want to increase the 17 V capacitor size. Don't
push the voltage rating.

Run the regulator and driver off the 25V. If there is an emitter follower
driver for the output transistors, run that off the 25V also. Your
regulator voltage drop will be equal to the VCEsat of the 2N3771, which is
1.4V max, .5V typ. Gain will be 40. As a comparison, the 2N3055 will have a
VCEsat of 3V and a gain of 15. The 2N5886 is marginally better than the
3771.

I have a schematic to my Astron supply, and that is precisely what they have
done.

Tam
 
L

legg

Jan 1, 1970
0
Albert,

I have looked at all the responses to this, and your other posting. Probably
the simplest solution is to use the brute force solution of adding a 6V 1A
transformer, as somebody suggested.

Or the separate lower-current diode/capacitor rectifier from the same
winding of the main transformer. Will give the same 17V but this is
loaded only be regulator and predrivers, so droop is avoided on the
regulator control and predriver supply, allowing the same low
saturation from pass elements without added magnetics.

If pass transistor beta is 10, control supply capacitance needs to be
larger than 1/7th the main capacitor's. If pass transistor beta is 40,
then it's capacitance needs to be larger than 1/30th of the main
capacitor's.

(snip)
Run your output transistors off the 17V. Change them to 2N3771 or better.
While you are at it, you may want to increase the 17 V capacitor size. Don't
push the voltage rating.

Run the regulator and driver off the

(snip)
regulator and driver supply.
If there is an emitter follower
driver for the output transistors, run that off the 25V also. Your
regulator voltage drop will be equal to the VCEsat of the 2N3771, which is
1.4V max, .5V typ. Gain will be 40. As a comparison, the 2N3055 will have a
VCEsat of 3V and a gain of 15. The 2N5886 is marginally better than the
3771.

I have a schematic to my

(snip)

GFC/Hammond supply
and that is precisely what they have
done.

RL
 
F

Fred Bloggs

Jan 1, 1970
0
Albert said:
OK, thanks to all who provided some guidance in the original post (see
'low voltage drop transistors for power supply' post.

Clearly, this was a 12v supply and not a 13.8v supply. The actual
output current ratings are not known, the only hint is that the
original meter was a 25 amp meter.

The transformer is much larger than 100 va and is actually larger than
a 200 va unit (based on core size). When I tore down the rest of the
assembly, I found more shaky stuff including sockets for the
transistors that were soldered, several etch cuts and lifted run on
the PCB, meter shunt was missing (probably discarded when the original
pcb was replaced). In other words, it probably needs a complete
redesign rather than using any of the original parts and pc boards.

I ran LTC's power supply design software and it came up with a 13.8
volt output switching supply that will supply 14A with the existing
power transformer. It uses the LTC1775 switching regulator, 2 X
Si4410DY mosfets, an 11 uH inductor and a couple of 1N5818 schottky
diodes. The supply itself (not including transformer losses and
rectifier losses) provides 98 percent efficiency at 14A output.

So, I'm considering building up this switching supply.

I need suggestions for a schottkey bridge rectifier that will handle
20A or more. I find the diodes themselves are readily available, but
I'd like to have a bridge assembly if possible.

Any suggestions??

Sounds like you don't have to change a thing- the 3055 is operated in
"inverted" mode and this will give you the lowest series voltage possible:
Please view in a fixed-width font such as Courier.

Inpn
----> Iload
3055 ----->
V+ --+------------ e c ------+-----+-> Vregulated
| \ / | |
| ---- Ipnp | |
| b ----> | |
| 2955 | Rbn | |
+----e c ----+----/\/\--+ |
| \ / |
/ ---- |
Rbp b error amp |
/ | /
\ | /| \
| | /+|------------> /
+------+-------< | \
\-|--Vref |
\| |
|
|
---
///



Light loading supplied by PNP 2N2955-

Rbn sized to cut-in NPN at heavy loading for boost-

At MAX Iload NPN is inverse saturated- very low VCE

Ie,npn=5A Ib,npn=5A -> Inpn=Ic,npn=10A

Ic,pnp=Ib,npn + Ipnp= 5A +10A= 15A @ Veb,npn =Low Pdiss

Iload/MAX=Ic,npn + Ipnp= 20A

Vbc,npn@5A - 12V
Rbn= ----------------
10A
 
F

Fred Bloggs

Jan 1, 1970
0
Fred said:
Sounds like you don't have to change a thing- the 3055 is operated in
"inverted" mode and this will give you the lowest series voltage possible:
Please view in a fixed-width font such as Courier.

Inpn
----> Iload
3055 ----->
V+ --+------------ e c ------+-----+-> Vregulated
| \ / | |
| ---- Ipnp | |
| b ----> | |
| 2955 | Rbn | |
+----e c ----+----/\/\--+ |
| \ / |
/ ---- |
Rbp b error amp |
/ | /
\ | /| \
| | /+|------------> /
+------+-------< | \
\-|--Vref |
\| |
|
|
---
///



Light loading supplied by PNP 2N2955-

Rbn sized to cut-in NPN at heavy loading for boost-

At MAX Iload NPN is inverse saturated- very low VCE

Ie,npn=5A Ib,npn=5A -> Inpn=Ic,npn=10A

Ic,pnp=Ib,npn + Ipnp= 5A +10A= 15A @ Veb,npn =Low Pdiss

Iload/MAX=Ic,npn + Ipnp= 20A

Vbc,npn@5A - 12V
Rbn= ----------------
10A

Make that last equation Rbn= (Vbc,npn@5A)/10A .....
 
Top