TomGeorge:
The voltage at A will always equal the voltage at B.
If you want to turn Q1 OFF, the voltage at C has to equal the voltage at B, that is the gate has to be at zero potential with respect to the source.
It seems I don't know enough to even ask the right questions. 
...or even ask them in a way that provokes useful answers.
Is anyone willing to suggest something between F1 and Q1 that would reliably get +10 to +18V on the far side of Q1?
Dirk
TomGeorge:
- The voltage at A will always equal the voltage at B.
Quite right, though that doesn't answer what I was trying to ask. See revised question 1) below.
TomGeorge:
2. If you want to turn Q1 OFF, the voltage at C has to equal the voltage at B, that is the gate has to be at zero potential with respect to the source.
Got it; thank you, Tom. I've revised the zener stack to provide just under 5V and use that to turn on a NPN - which in turn connects Q1's source with its gate:
Updated questions:
From what I understand of the 3KP16A specs, TVS1 starts shunting at just above 16V, but doesn't clamp completely until 26V. Does anyone with experience with TVS' know how TVS' actually behave? I.e. What voltage will Q1's source actually see if the power source goes north of 26V? (Thank you Tom, for providing me the opportunity to rephrase my question.)
The intent with R1, R2, ZD2 and ZD3 is that, at less than 18V, Q1's gate will be connected with ground. When voltage exceeds 18, ZD2 avalanches ("zeners"? - OK with everyone if I verb a noun? 
), the base of Q2 should see 4.7V, and Q2 connects Q1's source with its gate. Will this actually happen as I've described?
Should R3 maybe be moved to just after the junction of Q1's gate and Q2's emitter?
I can't seem to get the calculations right for all of the components in the "over-18V" part (R1-3, ZD2/3, Q2); am I even close? If not, where and/or what needs adjusting?
I am concerned that the over-20V shunted by TVS1 may get to Q1's gate, hence the presence of SD4. Is this (SD4) even needed?
I had asked: "Will the IRF4905 survive 5V at its gate while passing 20A for periods in the multiple of seconds? I've decided that under-voltage (under 10V) must be dealt with (and will be when we're done here), so this question is now moot.
Speaking of under-voltage protection: Anything under 10V should turn Q1 off. Could someone suggest how to go about this?
Assumptions that can be made:
This circuit will typically see +11 to +14.5V, but droops to 5V - though more likely 6V - can occur for multi-second periods
Current required will generally never exceed 20A
Power transients (below 0 and above 20V) can occur for up to 400ms
Reverse polarity protection is handled by SD1 (30A, 150V Schottky)
I appreciate any and all answers, comments, questions, etc.
Cheers!
Dirk
I seem to have a hit a significant barrier with trying to develop a circuit to manage automotive power. It seems I'm either asking the wrong questions, asking the right questions the wrong way, asking the wrong questions the wrong way; or no-one here knows answers to the questions I'm asking. One way or the other, I'm not making any headway.
So. I'm going to try a different approach: break the total goal into smaller segments.
1) Ensuring voltage is limited to a defined maximum
Outline:
Description:
Q1 BC327 [url=https://www.onsemi.com/pub/Collateral/BC327-D.PDF](spec sheet)[/url]
Q3 IRF4905 [url=https://www.infineon.com/dgdl/irf4905pbf.pdf?fileId=5546d462533600a4015355e329b1197e](spec sheet)[/url]
ZD1 1N4746A [url=http://www.vishay.com/docs/85816/1n4728a.pdf](spec sheet)[/url]
Questions:
Cheers!
Dirk
Hi,
Will the voltage in this circuit ever exceed 18V?
Ummmm.. you tell me. you are telling the story.. lol
How would I calculate the values of R1a and R1b to ensure ZD1 operates as expected?
This is where you need to get down and dirty, and build something. Zeners are not ideal so some circuit adjustments will need to be made.
What happens if less than 6V occurs?
The MOSFET will start to operate in its linear mode, its Drain to Source resistance will rise and current to your circuit will be limited and the MOSFET will begin to get HOT.
So you need to get down and dirty, and build something.
As I said in previous thread, too much thinking and not enough practice, if you try building some of this you will learn quicker than you are here.
A DMM is a very powerful tool.
Tom.... 
:o :o :o
Thank you for your comments, Tom.
TomGeorge:
Ummmm.. you tell me. you are telling the story.. lol
Hmmm. This then appears to be a case of asking the wrong question or the right question the wrong way. Maybe:
The stated purpose of this circuit is to ensure more than 18V does not occur. Does this circuit meet this purpose?
TomGeorge:
This is where you need to get down and dirty, and build something. Zeners are not ideal so some circuit adjustments will need to be made.
Another incorrect - or, in this case, maybe incomplete - question.
Add the assumption: The zener has been measured to breakdown at 18V.
TomGeorge:
The MOSFET will start to operate in its linear mode, its Drain to Source resistance will rise and current to your circuit will be limited and the MOSFET will begin to get HOT.
Thank you. A follow-up question:
What happens if Q1's source is connected to its gate at 10V and the source voltage goes to 6V or less?
TomGeorge:
So you need to get down and dirty, and build something.As I said in previous thread, too much thinking and not enough practice, if you try building some of this you will learn quicker than you are here.
A DMM is a very powerful tool.
90% of my purpose in these forums is to learn a bit about electronics. In this topic, I'm learning about power management - having a specific goal is helping. My approach tends to be to gain enough understanding so I know what I'm actually building on a breadboard. On this topic, I don't think I'm even close to that yet. With your - and others' - help, I'm getting there, though! 
Aside:
I actually have breadboarded a significantly different approach to this. Wasn't happy with it though, so here I am again. 
I am just a beginner with electronics. I have very limited resources: a hacked computer PS for 5V and 12V sources; very cheap DMM; a cap and inductance meter I have no idea how to use; a beginner's Arduino kit; misc components (a lot of which were purchased for a completely misguided approach to this same project); two, really cool LED desk lamps; 8) and very limited funds (so I'm somewhat hesitant to set fire to any more breadboards).
Cheers!
Dirk
Hi,
What happens if Q1's source is connected to its gate at 10V and the source voltage goes to 6V or less?
If you connect the source to the gate, the MOSFET will turn OFF, even if gate is at 10V, the gate and source will have the same potential.
Changing the source voltage while you have the gate connected to source will not change the OFF state of the MOSFET.
Tom....
TomGeorge:
Hi,If you connect the source to the gate, the MOSFET will turn OFF, even if gate is at 10V, the gate and source will have the same potential.
Changing the source voltage while you have the gate connected to source will not change the OFF state of the MOSFET.
Dude! You da man; I thought as much but it helps tons to have someone with experience corroborate my thinking. Thank you, Tom!
I'm thinking at the moment (by all means, correct me if I'm wrong) that R1a and R1b are likely workable at 1k each, but may need tweaking.
Next up, requiring a minimum of 10V. The following is exactly the same as the previous circuit with the addition of Q2 (a NPN vs the other's PNP), R2a, R2b, ZD2 (10V vs 18V), arranged exactly the same as - and in parallel to - the previous circuit.
2) Ensuring voltage is at least a defined minimum
Outline:
Changed: Description:
Q1 BC327 [url=https://www.onsemi.com/pub/Collateral/BC327-D.PDF](spec sheet)[/url]
Q2 BC337 [url=https://www.onsemi.com/pub/Collateral/BC337-D.PDF](spec sheet)[/url]
Q3 IRF4905 [url=https://www.infineon.com/dgdl/irf4905pbf.pdf?fileId=5546d462533600a4015355e329b1197e](spec sheet)[/url]
ZD1 1N4746A [url=http://www.vishay.com/docs/85816/1n4728a.pdf](spec sheet)[/url]
ZD2 1N4740A [url=http://www.vishay.com/docs/85816/1n4728a.pdf](spec sheet)[/url]
Questions:
Cheers!
Dirk
Hi,
When voltage is between 0 and 10V: both Q1's and Q2's bases are powered; Q2 then powers Q3's gate and turns Q3 OFF.
TomGeorge:
Hi,When voltage is between 0 and 10V: both Q1's and Q2's bases are powered; Q2 then powers Q3's gate and turns Q3 OFF.
Um, not sure what you were trying to say with this, Tom? (I had edited the wording on this before you posted your response.)
Cheers!
Dirk
Hi,
Sorry there was a bit of text after that, not sure where it went..
When voltage is between 0 and 10V: both Q1's and Q2's bases are powered; Q2 then powers Q3's gate and turns Q3 OFF.
When voltage is 10V, ZD1 will not conduct, so Q1 is turned OFF, because Q1 base/emitter current will be zero.
When voltage is 10V, ZD2 will not conduct, so Q2 is turned ON, because Q2 base/emitter current flow, Q3 will be turned OFF.
As voltage goes from 10 to 0V, the Q2 base/emitter current will fall, hence Q2 will enter linear mode and begin to apply gate voltage to Q3. (Practical Experimentation, will show if gate voltage low enough to make Q3 conduct.)
When voltage between 10 and 18V, ZD2 conducts, so Q2 ON, so Q3 gate and source at same potential and Q3 OFF.
Yes, both Q1 and Q2 will conduct and Q3 gate and source will be at same potential, so Q3 OFF when voltage >18V.
Tom....
TomGeorge:
When voltage is 10V, ZD1 will not conduct, so Q1 is turned OFF
[...]
Wasn't sure I could keep the logic straight in my own head. Happy to have it corroborated.
TomGeorge:
As voltage goes from 10 to 0V, the Q2 base/emitter current will fall, hence Q2 will enter linear mode and begin to apply gate voltage to Q3. (Practical Experimentation, will show if gate voltage low enough to make Q3 conduct.)
Yep, I suspected the 0-10V would be (possibly) problematic. You said earlier:
TomGeorge:
if you connect the source to the gate, the MOSFET will turn OFF, even if gate is at 10V, the gate and source will have the same potential.
Changing the source voltage while you have the gate connected to source will not change the OFF state of the MOSFET.
This says to me that, no matter the voltage (0 to +10V), if Q3's source and gate are the same, Q3 will stay OFF. Is this correct?
If the above is correct, the trick then is to ensure Q3's source and gate stay at the same potential while the source voltage drops - at least until the source voltage is low enough to no longer allow the IRF4905 to conduct at all or to not cause any damage if it is in its linear region. (I have no idea what this voltage would be, but suspect it's pretty low.)
My understanding of BJTs is not up to the task of determining when the BC337 will stop conducting enough to keep Q3's source and gate sufficiently close enough to each other to keep Q3 OFF. (I'm guessing that VBE and/or VCE are involved? ...and likely R2a and/or R2b will need adjusting to manage this?)
Um, help?
Cheers!
Dirk
Hi,
This says to me that, no matter the voltage (0 to +10V), if Q3's source and gate are the same, Q3 will stay OFF. Is this correct?
Yes.. correct..
Tom...
There's something wrong here...
It's related to your previous question on the voltages at A and B, where A and B were electrically identical.
If Q2 tries to do something different to Q1 then unlimited current flows between the two of them. Their shared output can not be at a different potential.
One of those (I'm not sure which) needs to have a resistor so that the other one can take priority and drive the Q3 gate as required.
Stop thinking about voltages. Think about where the current flows. If there's a fault condition, can you limit the current through your circuit until the power dissipation is low enough to not cook the components? Yes the input might be a zillion volts but if it goes through a 0.1Ohm resistor (i.e. the power wire leading into your box) then the current will be controlled.
TomGeorge:
Hi,Yes.. correct..
Thank you, Tom.
So, not answered yet is:
If the above is correct, the trick then is to ensure Q3's source and gate stay at the same potential while the source voltage drops - at least until the source voltage is low enough to no longer allow the IRF4905 to conduct at all or to not cause any damage if it is in its linear region. (I have no idea what this voltage would be, but suspect it's pretty low.)
My understanding of BJTs is not up to the task of determining when the BC337 will stop conducting enough to keep Q3's source and gate sufficiently close enough to each other to keep Q3 OFF. (I'm guessing that VBE and/or VCE are involved? ...and likely R2a and/or R2b will need adjusting to manage this?)
Cheers!
Dirk
MorganS:
There's something wrong here...
See? This is why I'm still here and still not (quite) ready for breadboarding; would have just had me another wienie roast (not that I don't like me a wienie roast on occasion).
MorganS:
If Q2 tries to do something different to Q1 then unlimited current flows between the two of them. Their shared output can not be at a different potential.One of those (I'm not sure which) needs to have a resistor so that the other one can take priority and drive the Q3 gate as required.
Yep, I did ask if Q1 and Q2 would interfere with each other in some way; they're never ON at the same time. I'm thinking a 220 between Q1's collector emitter and Q3's gate? (Yes, said in the form of a question.) Something a little smaller than the 330, like so:
My reasoning is that Q1 is a better place as its emitter is only forward-biased when we have more than 18V (and can "afford" a larger voltage drop), and smaller than the 330 so more current is available to charge Q3's gate. (I'm sure someone's going to jump in with exactly how poorly I described that...)
Or maybe better would be a Schottky on the output of both Q1 and Q2?
I am beginning to think that maybe current draw when voltage is nominal (between +10V and +18V) is going to be too large in this circuit. This voltage management circuit will be "on" (connected to the car's power system) all the time - including when the ignition is off. I was hoping to keep this "quiescent" current as small as possible. I believe it's currently about 54mA; I'd like to see 1/2 of this or less (there are more bits coming that may contribute to quiescent current draw). Of largest concern is when voltage is nominal; R1a, R1b, R2a and R3 are "in circuit" during this time. Theoretically, how large can these be before things start getting unpredictable?
For the same "quiescent current" concerns, would changing Q1 and Q2 to MOSFETs be worthwhile? This should remove R1a and R1b from current draw, reducing it by about 6mA. Hmmm... If increasing the resistor values can reduce current draw by 50% or more, then this isn't worthwhile.
MorganS:
Stop thinking about voltages. Think about where the current flows.
Ack! I know this is fundamental to electricity, but this circuit (likely like most) requires thought on both (BJTs are current switched; MOSFETs are voltage switched). I try to keep both in mind when considering a particular aspect...
MorganS:
Yes the input might be a zillion volts [...]
Maybe not for this project, but I'd like me one of these. 8)
Thank you for your comments, Morgan. Keep 'em coming.
Cheers!
Dirk
Before getting onto the next bit (reverse polarity protection), I thought I'd go over a couple of things and ask a couple of questions. Just so it's handy:
Circuit:
1) A few posts back, there may have been some confusion about how this circuit keeps voltage between 10 and 18. Here it is in logic table form:
10-18V window logic:
ZD1 Q1b Q1ce ZD2 Q2b Q2ca Q3g Q3sd
--- --- ---- --- --- ---- --- ----
0-10V | OFF ON OFF OFF ON ON ON OFF
[b]10-18V | OFF ON OFF ON OFF OFF OFF ON[/b]
18-26V | ON OFF ON ON OFF OFF ON OFF
2) Someone had a concern about what happens when voltage starts dropping below Q2's ability to keep Q3's source and gate potential near zero. This definitely is a concern, but it seems to me that Q1 will become more able to conduct when Q2 becomes less able to conduct. Wouldn't this help with keeping Q3's source and gate close to the same until voltage drops low enough to no longer be of concern at all?
Q1) I really do need help - at least guidance - as to what might happen when voltage starts getting too low to keep Q3's gate powered (and Q3 OFF). Is the way I have things now really going to work well enough in this situation?
Q2) Is this really as good a discreet circuit as can be managed to keep things between 10 and 18V? (I have a hard time believing I'm already that knowledgeable...) For example is R1c; though I'm loathe to add components, would using a diode here and at Q2's emitter be a better approach? Or how about higher resistances anywhere to help lower quiescent current draw? Any other tweaks anyone can think of?
PS It may seem I'm obsessing about the less-than-10V thing. However, voltage definitely can get well below 10V - low enough to have to do something to mitigate concerns about Q3 heating up because it could be in its linear zone for extended durations (battery dying). Over-heating Q3 is not acceptable. Everything must be done to ensure that doesn't happen. ...and I don't have the experience, knowledge or setup to test and/or verify this.
As always, I look forward to any and all contributions.
Cheers!
Dirk
I didn't get any feedback on my last post, so I'm hoping "no news is good news" and am forging ahead with some small changes, and will soon post the reverse polarity addition.
Here is the latest schematic:
Changes:
Note: I had thought the previous iteration would consume about 54mA while operating in its 10-18V window. Turns out, it was closer to 20mA. I believe this iteration is closer to 10mA, so definitely a step in the right direction.
I'm also unsure if replacing Q3 with a MOSFET (maybe a BS250) would be worthwhile (doing so would effectively eliminate the current consumed by R3a/R3b while under 18V). Anyone with ideas on this?
I'd be happy to be corrected on this, but I'm thinking replacing the BJT at Q4 with a MOSFET isn't possible because Q4 is required to be fully ON at voltages well below what MOSFET gate's typically want. Amirite?
Cheers!
Dirk
On to reverse polarity protection.
I'm loathe to use a diode due to the voltage drop incurred - even that of SBRs. Fortunately, a "reversed" p-channel MOSFET can behave like one (see this Infineon paper). Unfortunately, we need to protect against reversed polarity voltages in excess of what a p-channel MOSFET's gate can tolerate (-40V). So, stir in an 18V zener along with something to limit current through it, and we have:
Comments, concerns, et al are welcomed.
Cheers!
Dirk
The last bit for power management is the 26+V filtering. So, the power management bits for the main power feed for this project now looks like:
That said, I still am unsure and have questions about a number of things here:
There is no possible way I could have got this far without the amazing contributors in this forum. Thank you.
I'm working on linking up everything in this topic and will post that schematic once I've done so.
Cheers!
Dirk
Hi,
Have you built it yet?
Tom...
