Dump Load Controller Bd 1

[commanda]

Circuit in pdf format

pcb layout in Autotrax format

The complete unit, when finished, with all the options (clocked latch fet driver, fault alarms, timer for boost mode, battery temp shutoff of boost mode, etc) spans 5 pcb's.

Amanda

[willib]

it might help to put a string of diodes going from the 24V battery to the input of the regulator , to lower the input voltage .

because going from 24 to 5 v is a lot of potential , and may help it run cooler.

just a thought

[commanda]

I did have it running last week with all the leds alight, didn't notice any heat from the regulator. But I will explicitly measure the current and work out the dissipation in the regulator.

Amanda

[commanda]

I've now built & done basic testing of all 5 boards. Although I haven't completely wired my unit up, I feel confident enough to post the circuits & pcb layouts.

Component numbering now matches the boards, and the circuits should match the pcb's as built. Pcb files are in protel Autotrax format. Boards are all single sided, some with more links than others.

Commercial rights reserved. You can build one or more for yourself, the guy down the road, your dog, the guy down the road's dog, or even the pizza delivery lady. What you can't do is ship the design to China and make million$.

It can be built in stages, and you don't even need to build all 5 boards if you don't want all the features.

Although I have specifically targeted the design at 24 volts, it can be readily adapted to other voltages.

I've done all the interconnecting with ribbon cable, or 0.1 inch headers & plugs. It could be hard-wired, but I definitely suggest you follow my lead.

Photos of the completed unit in about a week.

Not a newbie project. You need to be able to make printed circuit boards, and be fairly competent at things electronic to build this.

BOARD 1.

Contains the LM3914 expanded scale voltmeter, driver transistors, and voltage regulator. The regulator is 5 volt, to suit the cmos logic on the other boards. It may need heatsinking. I've calibrated mine to read from 27-28 volts in normal mode, and 28-29 volts in boost mode. Selecting boost mode can be done with a simple switch, or from the timer board.  It is perfectly reasonable to build a minimalist system using just this board, calibrated to read say, 20-30 volts, and using just the top one or two outputs to turn on fets.


Circuit


Pcb

BOARD 2 (Latch).

Contains a 74HC374 latch, clocked by a 555. This means the fets turn on or off at 30 second intervals. With this, you can use things like pumps as dump loads, without fear of them turning on & off very rapidly,and not doing them any good.


Circuit


Pcb

BOARD 3 (Timer).

Plugs into Board 1 in place of the manual Normal/Boost switch. Has push-buttons for start & stop, and the timer will stop boost mode after a pre-determined time. I've set it to run from 15 seconds (for testing) to 8 hours. Steps are in binary(8hrs, 4hrs, 2hrs, etc). The board is laid out so you can run ribbon cable to a rotary switch, or just use a jumper to select a fixed time & leave it at that. There's also an accumulate mode, so that the timer will only run while the battery voltage is at or above at least the first led.

The battery temperature board also plugs into this, and will terminate boost mode if the battery temperature exceeds a preset level.


Circuit


Pcb

BOARD 4 (Battery Temperature)


Circuit


Pcb

BOARD 5 (Alarm)

The 74HC86 exclusive OR gates monitor the gate & drain of each fet, and raise an alarm if they are not as they should be. So an open circuit dump load or a short circuit fet will raise an alarm immediately. An open circuit fet won't raise an alarm until that particular fet is turned on. There is an audible alarm, and a mute switch. The alarm led wil operate as soon as the mute switch is operated, so you can't accidently silence the alarm without visual indication. There is also a connection from the 10th led on board 1 into the alarm circuit to function as an over-voltage warning. On a correctly sized system, this should never happen.


Circuit


Pcb

Amanda

[dinges]

Amanda,

Excellent job. Looks great, will have to study the design a bit further to completely understand how it works. But I'm seriously contemplating building it.

Probably on perf.board and for a 19" rack housing (have 2 of the same, which have been waiting for a good application; yours seems to be it). Your modular design and the modular case seem a good match.

BTW, I notice that in the 1st schematic the transistors don't have part.nrs. These could be any particular PNP universal transistor, I assume, but a complete parts list may be an idea for some.

Have still to study it properly, but am so far impressed. Before I'll commit to building I'll first await your first experiences (ab)using the machine.

(Sorry Ghurd, seems your project has just been put on hold  )

Peter.

[commanda]

Any general purpose PNP (eg BC557).

Also, should have pointed out; the main switching FET's need to be Logic Level types to connect to the output of the 74HC374.

In an earlier post, I did publish a circuit which outlined most of the logic on one pdf. This might help clarify what it's all about.

http://www.fieldlines.com/story/2006/6/1/84528/29208

Hardware wise, it could have been reduced to a micro (microprocessor; eg Picaxe) and a CPLD (Complex Programmable Logic Device). But that breaks one of my prime directives, ultra long-term maintainability. Programmable devices require source code, compiler, an OS to run it on, a programmer, a hardware platform to interface to the programmer, etc. We already have an OS which won't natively talk to the serial or parallel ports (I mostly use Linux, so you can guess which other OS I'm talking about), and hardware platforms with no parallel or serial ports. Who knows what the situation will be like 20 years from now. Right here and now, I can't even buy an eprom eraser.

And, a CPLD would almost certainly require a multi-layer board.

<end rant>

Amanda

[oztules]

Amanda,

You haven't left us much to say. It looks like you've nailed it.....

Will have to fire up the win98 partition to see the boards, but no doubt I'll be impressed with these too.

I rarely use multiple boards, as the interboard wiring up takes me as long as the board takes to make, but i understand your desire for modular design. Your discrete component design rather than a pic or similar makes this buildable by anyone with patience, and a little electronic background.      

I think pcb building is a mechanical skill.

Parts placement is a patience skill, and fair soldering is quickly learned.

If people can build a decent mill, then with the same dicipline they'll be able to build this/these too, as you've broken them up into bite size projects.

It is refreshing to see someone show others practical electronics..... (and no I dont have a need for a dump controller, but if I did this would be it)

well done......oztules

[dinges]

An eprom-eraser? Weren't you in Australia? Just put it in the sun (and sit next to it with a drink & a good book ?)

Have just started collecting the necessary parts. First in line will be the temp.sensor; This would be always useful, even for other tasks should I desire to.

3 potmeters on the board will be mounted so they can be adjusted from the front. LM385 instead of ICL8069; 100uA meter instead of 1mA one; and other tiny details. Nothing too major.

Also, I assume that with an extra driver for the FET one could use non-logic level ones? I.e. the cheap/standard IRF... stuff. Will look around a bit on the web for the details on such logic drivers. Just a transistor for level-shifting will probably do the trick too, I suppose.

Was at first going to make a remark on the LM3914 (obsolescence/age of it) but who cares, if it works. Glad you didn't go the micro-controller route, I hate them (though for complicated projects they may have an advantage). Ultra long term reliability & serviceability. I love it. Plus the fact that ALL the components I already have in my (well equipped) parts boxes. Guess that's what you get when you've been tearing apart electronic stuff for almost 25 years

BTW, if you have got any 'extensions' or extras in your mind or in the works, let us know; that way I could prepare for them while building.

Love it. Expect more builders will follow soon.

Peter.

[commanda]

Most of the interboard wiring, and terminating the leds to the boards, is done with ribbon cable. Crimp crimp, plug it in, done.

I also used resistor networks in several places to keep the component count down. It can be assembled with discrete resistors by pushing one end through the board, then joining all the loose ends together.

For the main Fets, I found some NDP6060L, 60volt, 48Amp, TO220, logic level gate from Soanar (Australia) for AU$1.66 each in 10 off quantities. IRF540N from the same people cost AU$2.67 ea in 10 off quantities.

Might do a diagram showing how it all goes together.

Amanda

[dinges]

Hi Amanda,

Have just more or less finished the first part of the controller: the temperature board. Just need to add the ICL8069 and the LM335. Built in a 19" rack system. Will post pictures in my diary when there's a bit more to show.

I'm planning on starting with the rest (probably first the alarm board; last in line will be board 1+2+timer, which will be on one PCB to reduce interconnects). I build on perf.board so I'm flexible

I still have a few questions (which will pop-up more often, if others start building it, I guess):

• function of S302?
  
• function of S301?
  
• alarmboard: where does R519 (10k) go to?
  
• function pot V1 & V2 (board 2)
  
• function LED301 (timer)
  
• J303 (timer); times out if temp>40deg. C?
  
• LED colours; should led1 be green, the last one red, and the in-betweens amber, just like in the 'other', earlier design of yours?
  
• HC86 & HC374. I thought CMOS could work on 12V? Only HCT needs 5V, I thought.
  
• I've added a 7812 at the battery temp.meter, so it has its own regulated power supply. 5V seems little to me.
  
• instead of a 2N7000, could a BS250 or BS170 be used? (much more common, over here)
  
• Z501...Z508 (alarm): which voltage are those zeners?
  
  

As you see, quite a few (easy to answer, I think) questions. One other point I have; do you see an easy way for me to add an external 24V battery charger. I.e., if there's no wind/sun, to use the grid to recharge the battery without the dumpload kicking in (i.e. a way to turn the battery charger off as voltage approaches right threshold. Was thinking myself of adding a transistor driver and relay at the LM3914 (first amber LED) so that as the first dumpload FET would switch on, it would switch the external power supply off.

Also wondering how big a load the controller presents to the battery in stand-by mode. But I'll find that out soon enough, I suppose.

All I can say is.... Wonderful. Isn't much you haven't thought about.

Peter.

[commanda]

Peter,

S301 & S302 are momentary action push-buttons. They start & stop the timer. U302A/B forms a flip-flop. When reset (stopped) pin 4 holds a reset on the 4040 counter. When started (set) pin 4 goes low, removing the reset from the 4040; pin 3 goes high, turning on L301 (to indicate that boost mode is on) and Q301. Q301 switches in an extra resistor in the voltage divider feeding the LM3914. This makes the LM3914 read higher, allowing the battery voltage to climb to a higher voltage before the dump load cuts in.

J303 powers the battery temperature board. The end 2 pins are power, the center pin connects to Q401. If the battery temperaure rises above the level set by P403, Q401 turns on applying a reset to the flip-flop.

V1 & V2 (on board 1) were meant to be test points to aid in calibrating the voltage scale.

Led colours; if using leds 2-9 to drive dump loads as I've built it, then I think my colour choice makes the most sense.

The battery temperature meter works just fine on 5 volts. The ICL8069 can be any voltage reference; I've even used leds at times to get a stable ref voltage. As long as the output of U401b is higher than the LM335 voltage at full scale (3.23 volts at 50 deg C) so P403 has enough volts. You can change R404 to get enough output at U401b.

74HCxxx only works on 5 volts. I do a lot of design with 4000 series cmos, and this caught me out badly.

Z501-508 are simply protection devices, and should be equal to Vcc+0.6 volts max. Cmos devices have internal protection diodes from their inputs to each supply, but won't handle much current. The zeners are there basically to protect these internal diodes, and should be used wherever any cmos device connects to the outside world. I used 5.6 volt, because I had them on hand; but 5.1 or even 4.7 would work as well. The circuit will work just fine without them, although more prone to failure.

BS250 is a P-channel device; not suitable.

BS170 looks alright.

External charger; hang a relay in parallel with your first dump load, and wire the charger through the NC contacts.

Idle current; Board 1 draws about 10mA (includes 7805), plus 3mA for each led. I'll re-measure it tomorrow with all the boards running. Other than the various leds, and the battery temp board, the rest is cmos & should draw only a minimal amount.

Amanda

[dinges]

Hi Amanda,

Work on the controller is progressing. Temperature board is nearly finished, just a few more connections to be made. Alarm board is rapidly progressing, am at about 2/3 now. See photos below for my implementation:

http://www.otherpower.com/images/scimages/3538/dumpload_rack_resized.jpg

(only the two boards with white stickers are currently in use; all other slots are still empty)

http://www.otherpower.com/images/scimages/3538/dumpload_alarm_and_timer_WIP_resized.jpg

http://www.otherpower.com/images/scimages/3538/dumpload_alarm_and_timer_front_resized.jpg

Will probably start this weekend on the major part, board 1+2+timer. Am out of perf board at the moment.

Still a few questions, if you don't mind:

• C301 (10u). In the drawing it says non-polarized, but I assume an elco is meant?
  
• what scale did you calibrate the temp meter (0-50 deg.C)? (i.e., what range does the meter instrument actually show in your case)
  
• timerboard: is reference made to Q201 (to board 2) but doesn't exist.
  
• p101 & p102; I assume they are for adjusting charge voltage & boost voltage, but please confirm.
  
• on timerboard (at J301) it still says +12V; I assume this should be +5V
  
• when building the timerboard, the manual boostswitch can (or: should?) be deleted? Think I won't implement this manual switch too (since I plan on the automated timer), but maybe there are cases where a manual override could be useful? Still, that could be done with the start/stop switches & flipflops, I assume, so no need anymore for the manual boost-switch?
  
• when one of the FETs triggers the alarm, does it sound the audible alarm as well? The way I understand it, only an overvoltage situation would trigger an audible alarm. If you could give a little bit more explanation about this part of the circuit I'd appreciate it.
  
  

BTW, you were right about 74HC... being 5V. I checked and was surprized. Also, quite a few times I thought I had spotted an 'error' in your design. And each time it turned out you were cleverer or had foreseen a situation I hadn't even thought about...

Did I say I was impressed ?  

Peter.

[commanda]

C301; I used a Tantalum, but non-polarised was not explicitly implied. Electrolytic can be used, but tend to become leaky with time, causing the timing to drift.

Temperature board; I calibrated to 0-50 deg C.

Timer board; the point which references Q201 is badly labelled. This is the input point for the accumulating timer. Ultimitely, it connects back to Q101. J302 is a 3 pin header. Use a jumper shunt to connect two pins together. Up is accumulating timer, down is straight timer.

P101/102 set the display range (27-28 volts in normal mode). Once set, this will become 28-29 volts when in boost mode. Use an adjustable power supply as a source to do the calibration. Start by adjusting the pots to the end of their travel such that V2 is at a minimum and V1 is at a maximum, then adjust them from there to get the desired display range.

All references to +12 are typos, should be +5.

The manual switch is obsolete if using the timer board. Once boost mode is started, it can be stopped manually at any time with the stop switch.

Alarms; overvoltage alarm is triggered by Q110 driving Q501. Fled509 will illuminate (I used a flashing led) and the buzzer will sound through S501. S501 is double-throw, not center-off as drawn. Must add a better component to my library.

The XOR gates, U501/502, monitor the gate & drain of each output fet. When the fet is off, the gate will be low, and the drain will be high via the dump load (and circuit breaker). When the fet turns on, the gate will be high, and the drain will be low. In both cases, the XOR condition is true, and the output of the gate will be high. In a fault condition, the XOR condition is no longer true, and the output of the gate goes low. This will turn on its indicator led (led501/508), and pull the collector of Q501 low via the diode (D501/508). This will activate the general fault led Fled509 and the audible alarm.

S501 will mute the audible alarm. It will also ground the collector of Q501 and activate Fled509 when muted, so if you leave it muted you will have visual indication of the fact, whether there is an alarm active or not.

Amanda

[commanda]

Dinges,

I just checked out your photos. When you said you had a 19 inch rack housing, I never thought you meant Eurocard.

I just uploaded an updated version of the timer circuit.

Amanda

[commanda]

Idle current, all 5 boards, no leds on, 11mA.

Interim photo.

Clockwise from top left;

Alarm board

Timer board

Temperature board

Board 1

Latch board

Have yet to wire up the main output fets and circuit breakers.

Amanda

[dinges]

Looks very nice.

I've been wondering for a while whether it's a good idea to add circuit breakers to a dumpload. Because just when you need the dumpload most it may open up. Then again, circuit breakers do help prevent fires...

What are your considerations for (not) adding them?

[commanda]

You've got fets which come on sequentially. If a dump load fails, it just moves to the next higher one when the battery charges a wee bit more.

Also, you've got both audible & visual indication of exactly which one has failed.

Why would you NOT fit circuit breakers?

Amanda