Linear PSU Power Stage

C

Costas Vlachos

Guest
Just to ask for some comments on the following power stage scheme for a
linear PSU:

http://www.btinternet.com/~cvlachos/psu-a.gif

Everything in the circuit is single-supply (no negative voltages anywhere,
and single-supply op-amps used). The BC556B/TIP126 pair provide a constant
1A current sink to aid programmability for capacitive loads. Due to the
single-supply operation, the current sinking circuit is only effective down
to about 2V. The BDV67A pair provide the high current sourcing (up to 4A),
and are mounted on the same heatsink for good current sharing. Load current
is sensed on the high side, a 1V/A signal is generated and fed back to the
LT1014 for current limiting (constant-current operation). For high frequency
compensation, there are 22pF and 270pF caps between (-) and OUT of U1A and
U1B, respectively (not shown).

So far it seems to work alright, but I haven't done extensive tests. What do
you think? Does it look OK? Any comments are appreciated.

Many thanks.

Costas
 
Costas Vlachos wrote:
Just to ask for some comments on the following power stage scheme for a
linear PSU:

http://www.btinternet.com/~cvlachos/psu-a.gif

Everything in the circuit is single-supply (no negative voltages anywhere,
and single-supply op-amps used). The BC556B/TIP126 pair provide a constant
1A current sink to aid programmability for capacitive loads. Due to the
single-supply operation, the current sinking circuit is only effective down
to about 2V. The BDV67A pair provide the high current sourcing (up to 4A),
and are mounted on the same heatsink for good current sharing. Load current
is sensed on the high side, a 1V/A signal is generated and fed back to the
LT1014 for current limiting (constant-current operation). For high frequency
compensation, there are 22pF and 270pF caps between (-) and OUT of U1A and
U1B, respectively (not shown).

So far it seems to work alright, but I haven't done extensive tests. What do
you think? Does it look OK? Any comments are appreciated.

Many thanks.

Costas
Click to expand...
The design seems dangerously close to the SOA limit at worst case
conditions for the BDV67A pair. Have you set the output to max voltage
out , input to max voltage in, current limit to max output current, and
shorted it. I have been bitten by SOA on darlingtons in the past as most
seem to be optimized for switching applications. Perhaps a third BDV67A
to give more SOA margin.
Good Luck
Jim Stockton
 
Costas Vlachos wrote:
"Jim Stockton" <mstech@pdq.net> wrote in message
news:3F12A356.9518EDDC@pdq.net...
Costas Vlachos wrote:

Just to ask for some comments on the following power stage scheme for a
linear PSU:

http://www.btinternet.com/~cvlachos/psu-a.gif


The design seems dangerously close to the SOA limit at worst case
conditions for the BDV67A pair. Have you set the output to max voltage
out , input to max voltage in, current limit to max output current, and
shorted it. I have been bitten by SOA on darlingtons in the past as most
seem to be optimized for switching applications. Perhaps a third BDV67A
to give more SOA margin.
Good Luck
Jim Stockton

Thanks for the comments Jim. Yes, I've been bitten by SOA too. That's why I
used two BDV67A in parallel. If you check the SOA graphs, you'll see that
*each* can take 4A at about Vce = 50V, so I assumed two would be enough.
Note that there seem to be many varieties of BDV67A around! The ones I'm
using are made by SEMELAB and are rated for 200W continuous power each. I've
found others made by MOSTEK (again BDV67A) but rated at 125W (?!?!?). The
data sheet for the SEMELAB ones is here:

http://www.btinternet.com/~cvlachos/bdv67.pdf

I think the SOA for them should be OK for my specs. Let me know if you still
think there is a problem.

On another issue, what concerns me is the BC546B/BC556B push-pull pair
(before the BDVs). Initially I didn't have them in the design, but had to
put them in for reasons relating to the limited current sinking ability of
the LT1014. I'm not very experienced in analogue design, so I wonder if this
"triple darlington" configuration has any dangerous pitfalls in terms of
speed/stability for the feedback loops. Anyone have any comments on that? It
seems to work OK, but I want to make sure.

Thanks again for the reply.

Costas
Click to expand...
The curve I looked at showed 2A at 40V Continuous.
Your part should be better.
Yes the triple darlington could have problems but there are resistors
internal to the BDV67A that should help. The degenerative feedback in
the emitters also helps with high frequencys. Perhaps others can comment
on stability, but I don't see anything glaring.
Good Luck
Jim Stockton
 
"Jim Stockton" <mstech@pdq.net> wrote in message
news:3F12A356.9518EDDC@pdq.net...
Costas Vlachos wrote:

Just to ask for some comments on the following power stage scheme for a
linear PSU:

http://www.btinternet.com/~cvlachos/psu-a.gif


The design seems dangerously close to the SOA limit at worst case
conditions for the BDV67A pair. Have you set the output to max voltage
out , input to max voltage in, current limit to max output current, and
shorted it. I have been bitten by SOA on darlingtons in the past as most
seem to be optimized for switching applications. Perhaps a third BDV67A
to give more SOA margin.
Good Luck
Jim Stockton
Click to expand...


Thanks for the comments Jim. Yes, I've been bitten by SOA too. That's why I
used two BDV67A in parallel. If you check the SOA graphs, you'll see that
*each* can take 4A at about Vce = 50V, so I assumed two would be enough.
Note that there seem to be many varieties of BDV67A around! The ones I'm
using are made by SEMELAB and are rated for 200W continuous power each. I've
found others made by MOSTEK (again BDV67A) but rated at 125W (?!?!?). The
data sheet for the SEMELAB ones is here:

http://www.btinternet.com/~cvlachos/bdv67.pdf

I think the SOA for them should be OK for my specs. Let me know if you still
think there is a problem.

On another issue, what concerns me is the BC546B/BC556B push-pull pair
(before the BDVs). Initially I didn't have them in the design, but had to
put them in for reasons relating to the limited current sinking ability of
the LT1014. I'm not very experienced in analogue design, so I wonder if this
"triple darlington" configuration has any dangerous pitfalls in terms of
speed/stability for the feedback loops. Anyone have any comments on that? It
seems to work OK, but I want to make sure.

Thanks again for the reply.

Costas
 
"Costas Vlachos" <c-X-vlachos@hot-X-mail.com> wrote in message
news:bet9p2$au6$1@sparta.btinternet.com...
Just to ask for some comments on the following power stage scheme for a
linear PSU:

http://www.btinternet.com/~cvlachos/psu-a.gif

Everything in the circuit is single-supply (no negative voltages anywhere,
and single-supply op-amps used). The BC556B/TIP126 pair provide a constant
1A current sink to aid programmability for capacitive loads. Due to the
single-supply operation, the current sinking circuit is only effective
down
to about 2V. The BDV67A pair provide the high current sourcing (up to 4A),
and are mounted on the same heatsink for good current sharing. Load
current
is sensed on the high side, a 1V/A signal is generated and fed back to the
LT1014 for current limiting (constant-current operation). For high
frequency
compensation, there are 22pF and 270pF caps between (-) and OUT of U1A and
U1B, respectively (not shown).

So far it seems to work alright, but I haven't done extensive tests. What
do
you think? Does it look OK? Any comments are appreciated.

Many thanks.

Costas
Click to expand...
Greetz }:)~~~

Thiz circuit is bumhole becoz yu hav not explaned it rather well. So I
cannot be boffered to do much more than throw it up in the bin wiv my over
puckle.

4tuneatlee, I am not implied buy the cumpanee you hav just bean sakd frum so
I dunt hav to figur out wut yu thogt yu wer doing n'ee'*way*. Perps you do
kno and kneed to disc the Manical.

Thanx

Greetz



---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.495 / Virus Database: 294 - Release Date: 6/30/03
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top