+24Vdc from +/-15Vdc
- Thread starter ioaudio
- Start date
Help Support GroupDIY Audio Forum:
peter purpose
Well-known member
onlymeeee said:
I crosed Dan Kennedy's palm with gold.
peter purpose
Well-known member
You only showed me a bit..!!
peter purpose
Well-known member
Well mine is bigger than yours.... ner ner ne ner ner.
Drum roll..... With massive thanks to Dan Kennedy and posted with permission.
Drum roll..... With massive thanks to Dan Kennedy and posted with permission.
peter purpose
Well-known member
Still got your name on it, but only if you explain the intricacies of the circuit.
peter purpose
Well-known member
Get on with it.... that Marconi is turning into a Sifam.
The turn on circuit will do.
The turn on circuit will do.
Hi Guys,
Just so you know, this circuit is in production, works, but is not perfect, or really
technically correct.
As an exercise for the class, what's the issue? (issues? might be more, but I won't pay
for any engineering
)
I had a few early warranty returns, the problem was identified, solved one way, but
can anyone tell me what they were?
Just so you know, this circuit is in production, works, but is not perfect, or really
technically correct.
As an exercise for the class, what's the issue? (issues? might be more, but I won't pay
for any engineering
)I had a few early warranty returns, the problem was identified, solved one way, but
can anyone tell me what they were?
> exercise for the class, what's the issue?
Not sure how it took the gang two whole pages-o-posts to fail to define the BASIC:
1) You want to wait for a good 30V total to arrive (not 0.6V or 4V). If you engage before BOTH (+16V and -16V) rails are fully "up", the cross-current can pull the weaker/slower regulator into reverse, and bust something.
2) You want to wait a little longer for other stuff to settle before you throw this heavy load on the rails.
Your 1N4750 is dead until 27V happens, then R1 R2 Q1 offsets another 1.2V before Q1 will engage. This delays firing from barely-started to nearly-ready (without much chance of failing to engage if the rails are a bit below nominal +/-16V).
As the rails pass 20V-30V, all the other cards in the rack are thumping, pulling charging current, generally coming to life un-gently. The 5K against 680uFd gives another few seconds of delay until Q2 can engage. Long enough for other loads to settle, not so long that the session is delayed.
> a few early warranty returns
Hmmmm. It does not dis-engage for several seconds after the rails slump. One diode or a re-shuffled resistor, or diode and added resistor, would change that. Ideally it would cut-out like cut-in: at ~~28V, as power starts to fail, not after it has quit.
However that would be hard on the rails, and you say "warranty returns", suggesting YOUR module failed. I'm always nervous about big caps across junctions, C15 across Q1. Depending what is inside "RC", it could discharge the cap through the base... but 2N4401 is a fat-base device and tolerates recklessness.
I'm not keen about the grounding topology as-drawn, but implementation and results trump intuition and theory.
Not sure. Not going to lose sleep over it.
There are voltage detectors for mini-brain systems which will drop a "Ready" line when 3.3V gets to 3.1V, to stage an orderly shut-down while there is still enough juice to think with. Ought to be able to shim that by factor of 9 to work here. But the datasheets and pinouts always baffle me.
Not sure how it took the gang two whole pages-o-posts to fail to define the BASIC:
1) You want to wait for a good 30V total to arrive (not 0.6V or 4V). If you engage before BOTH (+16V and -16V) rails are fully "up", the cross-current can pull the weaker/slower regulator into reverse, and bust something.
2) You want to wait a little longer for other stuff to settle before you throw this heavy load on the rails.
Your 1N4750 is dead until 27V happens, then R1 R2 Q1 offsets another 1.2V before Q1 will engage. This delays firing from barely-started to nearly-ready (without much chance of failing to engage if the rails are a bit below nominal +/-16V).
As the rails pass 20V-30V, all the other cards in the rack are thumping, pulling charging current, generally coming to life un-gently. The 5K against 680uFd gives another few seconds of delay until Q2 can engage. Long enough for other loads to settle, not so long that the session is delayed.
> a few early warranty returns
Hmmmm. It does not dis-engage for several seconds after the rails slump. One diode or a re-shuffled resistor, or diode and added resistor, would change that. Ideally it would cut-out like cut-in: at ~~28V, as power starts to fail, not after it has quit.
However that would be hard on the rails, and you say "warranty returns", suggesting YOUR module failed. I'm always nervous about big caps across junctions, C15 across Q1. Depending what is inside "RC", it could discharge the cap through the base... but 2N4401 is a fat-base device and tolerates recklessness.
I'm not keen about the grounding topology as-drawn, but implementation and results trump intuition and theory.
Not sure. Not going to lose sleep over it.
There are voltage detectors for mini-brain systems which will drop a "Ready" line when 3.3V gets to 3.1V, to stage an orderly shut-down while there is still enough juice to think with. Ought to be able to shim that by factor of 9 to work here. But the datasheets and pinouts always baffle me.
jdbakker
Well-known member
PRR said:
The potential danger with that approach is oscillation. It is not clear how much current the unit draws when it gets enabled, but if it's hungry enough it can briefly pull the rails below your trip point, 470uF caps or not. Then again a switcher presents a constant power load at its input, which is hardest on a (possibly current limited) rack regulator.
To the best of my knowledge the VPR spec says nothing about supply sequencing or minimum supply rail slew rate at turn-on, and I can imagine that this, with certain racks and certain loads, might lead to Interesting Situations.
PRR said:
Those are nice, and I believe they can still be had in TO-92, but you have to be somewhat careful not to exceed their output ratings in this scenario.
JDB.
[somewhat before my morning coffee, so I might well be off by a mile]
We use the reset parts in a lot of our micro controlled systems around here, PIC's are fond
of starting up brain scrambled and lose it easily if there is a momentary glitch in supplies. The
reset parts with watchdogs enabled are a standard circuit here.
They could be made to work here, but it's more partsy then this circuit.
No the problem is not with the turn on, it's actually in the supply filter circuit. The Q of the
inductor/capacitor legs is very high, the ceramic caps were failing. The solution was to go
to 100V parts, I had originally figured 50V parts would have a fair margin, but no.
The better solution is to redo the board, spoil the Q and engineer the filters rather than do it
empirically, but as PRR said "implementation and results trump intuition and theory",
Since the parts change there have been no failures, and there are a lot of these things working now.
of starting up brain scrambled and lose it easily if there is a momentary glitch in supplies. The
reset parts with watchdogs enabled are a standard circuit here.
They could be made to work here, but it's more partsy then this circuit.
No the problem is not with the turn on, it's actually in the supply filter circuit. The Q of the
inductor/capacitor legs is very high, the ceramic caps were failing. The solution was to go
to 100V parts, I had originally figured 50V parts would have a fair margin, but no.
The better solution is to redo the board, spoil the Q and engineer the filters rather than do it
empirically, but as PRR said "implementation and results trump intuition and theory",
Since the parts change there have been no failures, and there are a lot of these things working now.
PRR said:
The Low drop out regulator I'm using in a new battery powered design (TC1054/55) has an error output pin that switches when it loses regulation. Handy as a low battery indicator, or in a mains powered unit to bang the reset line. (sorry only avail in SMT).
I've experienced some issues with PICs getting real stupid at low voltages, and not being smart enough to just turn off.
JR
Sorry if someone has said this earlier (I didn't read every word in this thread).
In 60's and early 70's most European (transistorised) broadcast electronics worked with single 24V. During the 70's +/-15V became popular in mixing consoles but German manufacturers still continued with their single 24V. So it was quite common that a custom console included both 24V and +/-15V modules. That was not a problem since most German modules worked fine with 30V because they had a built in regulator. If not, regulator was added. With transformer balanced modules no modifications were required when interfacing these two types of modules. In unbalanced/electronically balanced case electrolytic capacitors were added to get rid of DC.
Are you absolutely sure that you need a switching regulator?
In 60's and early 70's most European (transistorised) broadcast electronics worked with single 24V. During the 70's +/-15V became popular in mixing consoles but German manufacturers still continued with their single 24V. So it was quite common that a custom console included both 24V and +/-15V modules. That was not a problem since most German modules worked fine with 30V because they had a built in regulator. If not, regulator was added. With transformer balanced modules no modifications were required when interfacing these two types of modules. In unbalanced/electronically balanced case electrolytic capacitors were added to get rid of DC.
Are you absolutely sure that you need a switching regulator?
edanderson
Well-known member
do you "need" a switching regulator? in general, as you point out, no. the original topic of this thread was for a design that was originally made for a single +24V supply, but the final design ran fine on bipolar +/- 16V, because it was transformer balanced and the 0V could float rather than being tied to PSU ground.
however, when this thread was re-opened, it was now about a neve circuit. assuming you want 24V rather than 32V, you could use a linear regulator to drop the 8 volts from 32 down to 24 (i believe this is what colin did with his version) but drop has to go somewhere... so you are making some heat. the efficiency on a switcher is probably high enough to make it a more power-friendly solution than a linear reg. with the 500 format you do have to be mindful of how much current your module pulls (dynamic and static).
i suppose it might be possible to make a neve type circuit run directly on 32V, but i'm guessing it would require a few changes to make it happy.
ed
however, when this thread was re-opened, it was now about a neve circuit. assuming you want 24V rather than 32V, you could use a linear regulator to drop the 8 volts from 32 down to 24 (i believe this is what colin did with his version) but drop has to go somewhere... so you are making some heat. the efficiency on a switcher is probably high enough to make it a more power-friendly solution than a linear reg. with the 500 format you do have to be mindful of how much current your module pulls (dynamic and static).
i suppose it might be possible to make a neve type circuit run directly on 32V, but i'm guessing it would require a few changes to make it happy.
ed
djn111
Well-known member
Hi,
I had some thoughts about the power supply too, but a good starting point is the exact specs of the power requierment of the 500 racks, does anybody have this, maximum rating, dynamic startup etc.
DJn
I had some thoughts about the power supply too, but a good starting point is the exact specs of the power requierment of the 500 racks, does anybody have this, maximum rating, dynamic startup etc.
DJn
peter purpose
Well-known member
Works like a treat Dan... many thanks.
Added a little more diode tax and it comes on at 14.5v.
Though the little bugger gets warm.
Added a little more diode tax and it comes on at 14.5v.
Though the little bugger gets warm.
Um, which little bugger?
In a normal MP-500NV the CC6 module is slightly (5 degrees C) above ambient.
This is with an average 100 ma load current in a burn in rack with good ventilation.
Nothing else in the circuit shows any heat.
In a normal MP-500NV the CC6 module is slightly (5 degrees C) above ambient.
This is with an average 100 ma load current in a burn in rack with good ventilation.
Nothing else in the circuit shows any heat.
