I'm trying to eliminate nine 78L05 regulators and replace them with one central regulator. I'm using the LP2975 which is an LDO driver.
Innards of the LP2975 (PIN OUT) Screen capture from the data sheet typical application.
formatting link
LP2975 DATA SHEET.
formatting link
I figure I can bring the IC ground out to the load and tie it to the main return with a blocking diode.The positve sense lead is tied to the cathode at the load. Schematic
I don't think the diodes go in series with the remote sense leads. What I have seen is diodes *inside* the power supply between the active lead and the remote sense lead. The idea is for the output to not go to VMAX if the remote sense wire breaks. Didn't look at the rest of it.
Diode D19 is there for reverse polarity protection. I have short cct and OVP on all nine loads. If there is a problem with one load I don't want it taking out everything with it. I'm using an LM358 for current sensing when it exceeds about 15 mA a P- channel disconnects the load by latching open. This also works for OVP because I have a 400W TVS there when it breaks down the 15mA will be exceeded and the load will be disconnected. I would upload another schematic but my connection for uploading craps out now.
There could be up to 100mA ( 9 * 12mA ) flowing through the return lead that is why I have the diode there (negative sense LP2975 GROUND) I don't want all that current flowing through a couple hundred feet of wire in the return sense lead. There could be up to a 0.56v drop (350ft 22 awg worst case) in the return line alone, that's what I'm trying to avoid.
I know this all seems small but it all adds up . I have an MMDL914T1G (0.3 TO 0.4V) drop diode at the entrance of each load for reverse polarity protection, I have a sense resistor for the LM358 Protection circuit 100mV. Then you add in line drops 350ft *2, my five volt regulator is now 3.7 to 4v by the time it hits my load.
That is why I originally used 78L05's on each load but they consume to much current. I've tested my LP2975 setup it only consumes a little over 400uA no load versus 9 times about 1mA for all the 78l05's.This is also cheaper one regulator.
If you have a voltage source at the load end, the D19 can make sense but D21 still doesn't. If the load end is a battery, you have a bigger problem if it is connected backwards.
Do you have multiple cables to the multiple loads? I think we may be able to come up with a better circuit with a little tinkering around. For example, a circuit that latches off can be made by running part of the control circuit on the the load side of the P channel pass element.
One reference can be used to control multiple regulators made from low power op-amps. This will save a huge amount of power if you have many loads.
I dont. I have some logic gates there.I dont think they like reverse polarity.
I'm using 4 conductor 22AWG. Changing it all out would be a pain,its buried in PVC conduits.Three conductors are already used. I've come up with a solution it would involve pulling another conducter through though.
I did some more testing this does work. I hooked it up on the bench and tested it to 90mA.I brought the Feedback pin of the LP2975 out to the load as well as the LP2975 ground pin. I placed 4.7 ohms resistors in the forward and return path to simulate a 300ft run of 22AWG. R34 and R33.
This is the internal schematic of the LP2975.The one I posted earlier is the wrong link sorry.
formatting link
This is my test schematic. With the 4.7 ohm resistors.
formatting link
I'm beginning to think this is a PITA. What I might do is just leave the CDSU101A in the feedback path, this gives between 5.52V to 6.68v. The cds101a only varies about 90mv over the operating temp range for the small 50uA current in the feedback branch. This would give enough headroom for about a 170ft run or 340ft including the return. Replace the polarity protection diodes on the load with 0.22V schottkys.
This is probably what I'll wind up doing using the CdS101a to bump the output voltage up about 0.6V. It will limit the length of the run to
As with any remote sensing configuration, you can only hope to regulate across one branch with one regulator.
This remote regulation can only be effective at low frequencies. Decoupling at the load has to deal with load transients. The local resistive connection to the rails at the regulator, bypassed by capacitors, allows the regulator to maintain local stability at the higher frequencies, while substituting for the LF connection in the event of a sense line going open circuit, or in your case, S1 opening. Ideally the sense lines are also fitted with impedances that prevent damage under single fault abnormals, though fusing is unlikely at these current levels.
What kind of load fault are you expecting to result in reverse polarity? Charging circuits might reasonably expect it, but distributed supply regulators are more likely to experience the usual opens, shorts and static discharges of a communications buss.
The loads are in paralrel I planned on placing the remote sense leads on the furthest load.As an example the first load could be 200 to 300 ft from the regulator,and then 25ft between each load.The load can operate from 4.6 to 5.3V. Absoulte max rateing 7V.
The loads are small and I want them easy to take on /off move around,so there is a possibility of inadvertently hooking the wires up in reverse polarity. Worth a few cents for a diode a 20 V shottky only has 0.2V drop at 25 deg C and 12mA.
These are just IR RCVR's a project I've been working on improveing over the summer when time permits. My real job keeps getting in the way:)
The current to the loads is DC I used a ZXCT1009 to observe it on a scope. The 38kHz pulses are supplied locally from a 10uf ceramic mounted at the input terminals.This cap is replenshed with the average current (dc) from the regulator.
I thought I would at least investigate the possibility of remote sensing. I need a 5V regulator to power up an SSR,so I figured there must be a way to use the one regulator to power all the units. Placeing 78L05 on each module seemed a bit excessive,as well as the power they consume just running.
I've measured the NL current consumption of my LP2975 setup with my battery switch out circuit and tlvh431 refrence and it is only a little over 600uA, which is a far cry from even one 78l05 which can be several mA let alone 6 or 10.
The cheapsest and simplest solution I guess is to leave the CDSU101A in the feedback path,this puts the regulators O/P at the terminals at
5.5 to 6.6v (depending on temperature). This gives enough headroom for
170ft.This would ensure that even the nearest load dosent exceed 5.3v and the farthest load dosent go below 4.8v.
There seems to be some confusion over what is planned, and what currently is installed or breadboarded. Frankly, it looks like you're throwing parts at a blank schematic, hoping they'll somehow think for themselves.
It's uncommon to power long-wire, built-in hardware infrastructures permanently with a battery source. It's more common as a back-up strategy. Why has this decision been made? This is obviously not a portable installation.
If you knew me you would realize nothing could be further from the truth. I've seen on numerous occasions what happens when people do that it usually results in smoke and frustration.
There is nothing on that schematic that is unnecessary. On the contrary some components are multitasking to save space and power.
I've achieved all the goals I set out for when I decided to do this that being a low no load power consumption (well under 1% of rated full load), and a small PCB (1.3" x 1.6"),for a small outlay of cash for components. It could be smaller but I'm limited here by my manufacturing capabilities.
It worked as intended on first power-up, adjustment of the base resistor on the EMC2DXV5T1 to set the OVP to 6.5V was all that needed to be done.
The loads are not powered by a battery (I do plan on building a UPS for this though). D3 is where a 12V Flyback enters and takes over at about 11V, and then the battery is removed from the circuit (U7, D21).All loads are blocked by the N -chan mosfet in the LDO'S output until the flyback takesover.
The divider R22 and R7 (with hysteresis) sense the flybacks rising voltage, at about 11V U7's gate is pulled high. The other comparator is used both to sense a low battery condition (6.7V) as well as a normal user generated shutdown (the switch is swung open).A switch is at the input J2. This is where the battery enters,its sole purpose is to provide intial power to a bidirectional SSR that controls the Flyback.
When the switch is open the drain of U7 is at 0V therefore the gate of the nts4001 (U19) is at zero, this puts the comparator input at zero, the output swings high and fires a latch. I didn't use hysteresis for the second comparator because it fires a dual (NPN/PNP) U22A configured as a latch (SCR) with a 10nf cap for transient immunity. The resistors would just be a waste of space. The 4.7M resistor discharges the nts4001 in about 32Us fast enough for this.
The latch is wired in a way that it will shutdown the LP2975, This will shut everything down including the flyback.
I'm sorry the reason I initially only posted the portion of the schematic that was relevant to my particular questions, was for clarity for the specific question (remote sensing). The full schematic would have just been a distraction. The Additional questions were the reason I posted the full schematic.
It's not easy addressing unfamiliar schematics, particularly if they're a little dim around the component ID and labelling.
Bells and whistles are always a headache to unravel. It's made no easier where dual transistors are reduced to blocks, with pin numbers not obviously related to function - though I suppose it's no worse than an unfamiliar IC, similarly labelled.
Seeing this, I naturally tend to get more than a little doubtful about my ability to read the schematic accurately enough to offer useful input.
The circuit appears to run off the source at J1 by preference, without any provision for charging a back-up source through J2. Running off a battery arbitrarily through J1 wouldn't make sense.
A low-power reference could run off the 10V rail through a 100K resistor, without being fed from a constant current. A wide supply compliance is supposed to be one of the features of a voltage reference. There appears to be no appreciable load on the reference node requiring extra bias.
The signal latch around U22A is triggered by either the input( ?? at J2 ) or an output overvoltage. It is reset when both input terminals are opened. The latched signal goes elsewhere, so effectiveness in protection or ease of reset can't be assessed. The soft shutdown of the regulator through pin#1, in the event of an overvoltage situation caused by reglator failure, can obviously not fly.
The two parallel output lines are one more than the remote sensing method previously discussed could handle.
Yes I am sorry about that I tried to get it clearer.
I originally did have CI, C2, and B1 etc. On those duals but layout was being a PITA so I switched just to numbers.
I uploaded the schematic to abse in pdf format. I 'm in the process of transferring it to a dual sided board if you do notice something that could be improved I'm open to suggestions. Subject LP2975 (sed)
The current source is there (U23A) to provide a stable worst case
100uA to the TLVH431, I'm also using it to supply just the current required to the shutdown pin. Tieing the reference to the rail to a resistor wouldn't be as reliable and would draw more current then necessary. This keeps the reference properly biased down to were the tlc3702 shuts-off 3V. The current source and reference perform double duty it sinks just enough current into the LP2975's shutdown pin to guarantee operation over its temp range. Any small excess depending on the temperature goes to reference.
By placing an NTS4001 between the reference and the LP2975 shutdown pin I have a cheap and simple method to shutdown the LP2975, whenever the latch is triggered. By shutting down the LP2975 this shuts everything off flyback included.
The reference also is used for the dual comparator.
I'm not sure I follow you here. I'm guessing you mean that the protection shutdown wouldn't operate if the LP2975 were to fail this is the most unlikely scenario. Designing protection for every possible component failure possible would be a bit excessive. Instead the most likely would be failure of the pass FET.
And I did test the remote sense as I mentioned earlier,by placing a
4.7 ohm resistor in the positive sense lead and a 4.7ohm resistor in the IC ground lead. I tied the IC ground to the load being sensed and yes it does work. The reason I elected not to use it is because it would require another wire.
Further description:
J4 is for protection circuits I have on the flyback (thermal and OVP) if they activate they'll trigger U22A shutting everything down.
The second TLVH431 is just being used as an inexpensive comparator, this keeps U16 (FET) blocking, until the LDO's output cap charges up. I found this was necessary to guarantee predictable operation of the SSR which is fed through J5. A signal already exists, so I used an SMD
2n3906 to disconnect U14 once it's served its purpose.
U12 blocks the loads until the flyback takes over. I don't want the battery supplying the loads so a cheap Si1302DL (90mV worst case drop) is used to block the loads and an existing signal is used to gate it at the appropriate time.
The time that the battery actually supplies this circuit including
Would a soft shutdown of U2 provide overvoltage protection in that event?
There are basically two options in overvoltage protection:
- employ an independant series switch to withdraw power from the regulator. With redundant input power, soft or hard inhibition of both would be neccessary.
- employ an independant shunt switch to crowbar the point of overvoltage. This relies on the idependence/functionality of current limiters, or the presence of fusible series elements. If the voltage regulator was also responsible for current limiting, it cannot serve this purpose.
It's funny how discussions on the likelihood of one failure mode occuring vs another seem to take place only after a selected method of protection is identified as being ineffective. I'd take this a few steps farther back: Is the load threatened by unrecoverable damage in the presence of the fault? If not, don't bother. If so, then effective protection may be economically justified. If justified, then reliable protection is warranted regardless of the failure mode. Waffling over the issue is probably a waste of time.
One thing that an OVP latch will benefit from is a simple method of reset - useful in testing and in troubleshooting. This may as simple as two pads that are easily shorted by a screw driver.
An admirable first step. Don't you think there may be other characteristics exhibited by a 200ft-long wire harness that might warrant further investigation? If the hardware that was proven by the previous circuit incarnation is still present, it could be a simple matter to use it as a breadboard element.
If the signal at 'delay' can be used to disconnect the loads, it may be a suitable series element for output OVP protection.
In most redundant source systems, there comes a time when you have to ask yourself 'Who's on first?'. There may be more than one answer, but all must be anticipated.
If you're attempting to imhibit nodes in a protection scheme, you don't want your protection circuitry to be dependant on those nodes for continuous maintenance of inhibition. Hence the next question - 'Who's off last?'.
Yes it would. I'm sensing the LDO's output when it goes low it will disable drive to the SSR, and disconnect the flyback from the AC line. So anytime that latch is fired disabling the LP2975 it will shutdown completely. The SSR requires no power to shutdown in fact a loss of power will shut it down.
Yes the LDO does use current limiting but its not driving the the latch it's driven from the flyback. If the ldo does go into current limit the voltage would collaspe (Short cct.) and diasble the SSR.
Yes assuming the LDO fails and loses regulation (for whatever reason) the maximum voltage that the loads on the LDO can take is 7V. So yes they would be destroyed.The loads all have 400w 5v TVS diodes but these couldnt be reiied on for protection. They are just there to clamp any transients when the loads are turned on/off.
then reliable
Agreed
Yes I'm not transmitting data at Rf frequencies.R dominates at DC which is what is going through those conductors.Voltage transients can be generated by the inductance at turn on/off which is why each load has a TVS diode on it.No need to break out the smith charts.
Your independent protection system requires power,what if this fails? Where would it stop by this rationale you should also provide monitoring and protection to your protection circuits and then monitoring and protection to those ......;)
The first step is to try to reduce the likely hood of a failure/fault by minimizing the stress on components. The goal unless your building something where lives are at stake is to try and anticipate the most likely point of failure and then build a suitable protection network.
For milatary aerospace yes if they want to pay for redundant backups fine. For most application what I'm doing is significantly more then what I've seen.
I will be testing it under as many fault conditions that come to mind if adjustments are needed they will be done.
You have given me some addtional food for thought,thank you.
It is, of course, your intention to supply DC to the load. At the same time, however, you also intend to regulate the voltage at this remote branch. The act of regulation is attempted using the combined functions of the integrated circuit, the pmos pass transistor, the output filter and the feedback network.
The typical application circuit offered by the mfr does not assume the presence of a 200ft harness, nor does it expect remote connections on it's ground or sensing terminals. The onus is therefor on the end user to research and demonstrate functionality. Is this not the origin of the "Would this work" subject line of this thread?
The line impedances will not be accurately simulated by two 4.7 ohm resistors for a few simple reasons;
-There are more than two loops of wire here - there are at least three.
-The resistive, self-inductive and capacitive characteristics of the four-wire harness produce differing effects on each functional loop - aggravated by the intentionally low capacitive impedance of the regulated branch that appears in the midpoint of the loop and the fact that the lines are shorted by direct contact and indirectly through this impedance. This shunting impedance is repeated in known fixed increments as the nimber of load circuits is varied.
Granted; the first-order approximation of the inductance of 300V UL styled 22AWG stranded cable does not produce a intimidating number - roughly 200nH per foot of a 2-wire return ( minimal area loop). For twisted wire pairs this could be reduced. Given an available fixed installation, I'd be tempted to make an actual measurement, to know unambiguously what the passive characteristic values actually were. Designing for a fixed lines or known incremental multiples of loading can actually be simpler than designing for 'universal' applications where the loading is unpredictable.
Even so, the linear controller is not expecting the LCR terms to appear - it only anticipates the first-order delay through the pmos signal inverting pass element and (likely) expects that the local decoupling cap will dominate.
The concepts of 'fail-safe' operation and 'fail-safe' fault signalling are ones that assumes that the simplest and most likely failures will occur first with the most likely effect (further simplified here as open or short). The order of failure can be assumed to be related to the established failure rates of the components involved. Some investigation into this aspect of component reliability in design is worthwhile - it may offer illumination, as it will either confirm or modify prejudices formed previously from various sources.
The 'operating' end of the issue will generally dictate good margins on design parameters, combined with parallel or series redundancy. Hence the very existence of battery-back-up in your circuit - as an example of how two devices with relatively poor reliability might be configured to assist in providing marginally improved security of function. A common and perhaps more extreme example of a similar situation is present when short-lived cooling hardware is employed in miniaturized or over-stressed power electronics, in the ubiquitous PC.
The 'fault signalling issue' is resolved by attempting to configure circuits so that the most likely failure mode will allow targeted priorities to be maintained. If the priority is continuous operation, the failure will not result in discontinuity of operation. If the priority is safety, the failure will more likely result in an inhibition. If priorities are not established, a dissatisfied end-user is likely the first, but by no means the most serious, eventuality.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.