Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Help with a Delay Relay circuit

Not open for further replies.


New Member
Hi, Let me start off that I know very little about how to design circuits, but am confident I will be able to put together a design if one is handed to me. The reason I'm posting here is I have a computer system I built for myself but I have found that the video card I am going to use (ATI AIW Radeon 9700pro) draws too many amps from my PSU on the auxillary power connector that the card uses when I press the power button. I have also discovered that if I can find a way to delay the ATI card from recieving power from its auxillary connector for just 1 second after I press the power button it will not trigger the overcurrent protection in my PSU and thus the computer will boot properly and run just fine. This is an issue that has been documented by many at various hardware enthusiast forums and alas we're good at building computer rigs, but few of us know that much about designing simple circuitry devices such as a Delay Relay.
So I hope that a kind soul here who happens to be good at this might be able to share with me a simple way to design a delay relay device that has a failsafe (so if an electronic part of it goes out it won't fry other expensive pieces of computer equipment) and takes input from 12V and 5V (the power that is sent through the 4-pin molex connectors on a computers PSU). I'm certain that given a design that will produces the desired delay of 1 second til power reaches the video card, I can have a working computer.
Thanks in advance for any help you might offer
Sounds like a job for a 555 timer controlled relay. You can buy kits for that.

**broken link removed**
Here this should work, actually i know the timer delaying will work because i just built it and tested it :) . Im not sure how sensitive you power supply is because you might have a problem with this design due to delay times (mainly the relay). Ill test it some more tomorrow about the relay thing. If its is too long I can use a different design (not the normally closed contacts). I would prefer this design though because the relay is off except for that one second, I don’t like the idea of leaving the relay on the whole time the computer is on, plus if it fails somehow you are screwed. That’s why I like this design better. Based on the old monostable 555. The only other problem i could see is that the cap used to trigger it could not discharge fully, but this should really never be a problem since it discharges in about 5 seconds, just dont boot then shut down and reboot within 5 seconds.
Simple operation: When first powered up the relay flips (disconnecting power to the card) for 1.1 seconds, then the relay shuts off connecting its normally closed contacts again, thus power is delivered to the card.
Hope that helps some.


  • Delay.gif
    7.7 KB · Views: 2,023
That looks really nice, and I appreciate very much you coming up with that design for me. I do have one question though. From what I understand, the video card takes the 12V line and dumps it (probably to ground) and actually powers itself off the +5V line. That said, I still was wanting to include all lines that the normal floppy connector had just to make sure everything that the designers of the card planned to have going to the card was. So the card gets both a +5V and a +12V and the 2 ground lines from the 4 pin floppy connector that hooks into it. So my question really is, I didn't see a +5 Volt input to the relay or a +5 volt output, so I was wondering if that was accounted for in your design which looks to be a good one aside from the +5Volt line thing.
Thank you again for posting that design though and any further help (or explanation if I'm just not seeing something thats already there) would be much welcomed e.
yea i see what you are saying, i would do that too (isolate both the 5V and the 12V). I revised the circuit. I just built the whole thing and it seems to function nicely on the breadboard.
Upon power-up it holds the relay on for 1 second then turns it off, that would then supply power to the card. I was looking on the net for some specifications on the current draw of that card but came up with nothing. I did see a lot of people complaining about this exact problem though. Most say it seems to be a problem with the power supply being triggered only at start-up, so this idea might work. The only thing is im not sure if the system will get past the POST...? Im not sure if you tested this yet though. If i were you i would hook a switch inline and turn the computer on then flip the switch a second later simulating this circuit, that way if it doesnt work you would not have wasted any money building the circuit (although this will proabably cost about $5 to make lol)

I tested this using a 12V SPDT automotive relay cause it was the closest one around, the coil resistance is 70 ohms, the circuit requires a DPDT relay, try to get one with a high resistance (>100 ohms). That will lower overall current draw.

Tested this circuit draws about 135 mA during the first second (~1.7 Watts), and about 6 mA after that (~72 mW). Note that this will draw the most current when first powered mainly from the relay. I dont think this is enough to cause the same overcurrent protection problems.

Also for building this circuit it might be a good idea to get a male and female molex the same kind that way you can just plug it inline and dont have to cut any existing wires.

One concern with this circuit: If you shut down the computer and restart within about 5 seconds it will not do the delay thing, although i would not recommend turning a system off then on that fast anyway. This would probably mean that using the standard "restart" function will not work. You will probably have to shut off then on at least 5 seconds later. Although i dont think these are major concerns.

Anyway here is the revised circuit, if you have any questions feel free to ask.


  • Delay_162.gif
    11.7 KB · Views: 1,793
I recieved a better explanation of what is going on that causes the PSU to shut off power to the system and why the 1 second delay works. Another person who I believe put together some type of delay/relay circuit as you have shown me said something about the video card sends the +5V line from the auxillary connector down the 3.3V line that is present on the AGP slot on the motherboard. It might be doing this to change the voltage that the card is able to use, I really don't know though. Anyway when it does this on power up, the PSU (and particularly the ones I have) will power up the 12V line first, then the 5V line, and finally the 3.3V line. which means the initially when the PSU powers on it powers the 5V line and before it can power on the 3.3V line it recieves 5V current coming back to it on the 3.3V line. This triggers a protection circuit that would shut off the PSU to avoid damaging itself. At least that was the latest description of the reason this is all happening. I suppose it could be that, I really don't know, to me it sounds about as good as an overamping problem, but in anycase, the 1 second delay works and solves the problem, and a solution to this problem is enough for me to be happy.

Thank you very very much for your help,
I will get the components you specified from a local electronics shop and some breadboard and start making this circuit. I'm sure I can test it with a multimeter on a spare (read expendable) PSU before I attempt using it on anything expensive. I was planning exactly as you said to get 2 molex connetors to make this more plug and play. In fact, I'm planning on either mounting the design to a PCI holder (so that it will fit verticle in a PCI slot, or I will mount it just underneath the floppy drive where I should have enough structural metal to screw it on. Thank you again for your help. I plan on sharing this with all who have this problem in the various forums I've been posting on, and of course I will give credit directly to you and link to this thread.
I don't know if I can thank you enough really this has been such a headache for so many people and on a $400 video card which makes it even worse.
hey no problem, sounds like this should do the trick then, since the 12V comes on first that makes it even better. Let me know how it turns out. all those components are very easy to find since they are all common. You post on a forum where this is a concern? shoot me a link and ill write an explanation of this thing.
Ok I have put together a list of various places I've posted about this problem and other people have posted about this problem. My name on all these forums is CVNet1. The 2 forums that had the most response about this problem are the and the rage3d forums. I'll post all the links to each one (an asterick by the ones that will probably be most effective to post your solution in), and you can feel free to post to anyone of them. Thanks again.

**broken link removed**


**broken link removed**

* **broken link removed**

* **broken link removed**

**broken link removed**

I'm quite fond of relay type answers.
Especially for short time intervals like this.

But i don't like this one,
cos the relay starts with those contacts closed.

The same supply that works the relay goes to the card.
So maybe the card gets a feed, then its gone, then
its back when the relay relaxes.

I would try to use SCRs instead,
They would start by being OFF,
and they are easy to work.

a resistor and cap on the gate would give a short delay.

Try to keep the cap fairly small, as when it operates the
SCR the cap discharges through the gate.
If you need to make the cap more than 6 mfd (thats a guess)
then put a few hundred ohms in the gate, just to limit it.

I don't know what value res would give one second,
i would try this outside the PC working a small lamp,
to get the values about right first.

I would guess at about 30 k ohms.
Others may be able to give closer figures, but it also
depends on the SCR you get, or may have to do it.

It may be possible to simply disconnect the 12v feed.

I'm not familiar with the card you speak of,
i looked through all the links you posted
to see if anyone else had posted circuits or diagrams,
but i couldn't see any.

If this does not suit you just ignore it ...

Best of luck with it, John :)

(i am running 640-480 and i am fed up panning left and right.)


  • SCRdelay.jpg
    9.3 KB · Views: 1,647
john1 is right an SCR is better suited. I just replaced the relay with an SCR in the board and it works great, I dont know if i would just use a resistor and cap and base it on charge time, but you could try it, i dont think the timing would be as accurate... I used a C106D SCR, (closest one around), and it works but it will only handle 4 A, i dont know if that is enough but you can always get one that will handle more current.
Good thinkin john1,(sorry for the high res, i didnt know people used 640x480 for anything other that safe mode :lol: ), but small diagrams are hell to read.

I redrew it for you:


  • Delay_152.gif
    10.5 KB · Views: 1,633
Thank you 'e',

Yes, small diagrams are a pain, though they're ok on this
10 and a half-inch wide screen.

I see you have put SCRs on both the 5 and 12 volt lines,
i couldn't find much about this particular card, but i
got the impression that it may run ok without the 12 volt
connected at all.

If it wont run without the 12v feed on as well, you could
put another SCR in the 12v line with its gate fed from
the 5v delivery side.

The delay from a resistor to the gate with a cap is quite
ok, its used in many things.
And i would be happier with a limiter on that gate, maybe
5 k or so.

Sorry about the tiny drawing, but its not so tiny for me !

Most people have small monitors, its only wealthy
people that can afford nice big displays !

Regards, John :)

just press return after about 15 words, then anyone in
the world can read your post easily.


  • 2xSCRdelay.jpg
    9.2 KB · Views: 1,572
well the reason i am weary of using a cap and resistor is because the SCRs only need 15uA to trigger the gate,
that means you will need a very large time constant to keep it off for one second.
What do you mean "limiter" on the gate? The transistor sinks
the SCRs to ground for the first second.
It then becomes unbiased and allows the node to go up in voltage
via the 220K resistor allowing around 50uA, enough to trigger the
SCRs gate. I dont see the point to limiting the ground potential,
unless you are on to something i dont see.
Yea typing with the breaks makes it easier to read, well with 1280x1024
its fine as is :lol:
Hello 'e' ,

Well, when i mentioned a limiting res on the gate, i was
actually referring to the diagram that i had drawn.

I had not looked at yours very closely, i assumed that it
was workable in all its major points.

I've increased my settings to 800 x 600 just so as i can
get most of your diagram on my screen without panning.

My monitor (Dell Vi 428 EBP) does not take kindly to my
video driver at that density, and reduces visible area
by about a quarter of an inch all round. Most unfriendly.
Especially on this little monitor!

However, ive had a good look at your diagram now.
Yes, i can see your intention for the transistor Q1.
It seems that you want its output to be at a very low
potential, to keep both SCRs switched OFF, until the 555
tells it otherwise.

Most SCRs are pretty fast, i would have to look up the
details to know just how fast, but i think they would
give a 2N2222 a run for its money.

See where i'm going here? ... that 2222 has to get its
instruction to conduct, and keep its collector voltage
nice and low, from the 555, before those SCRs can
react, cos once they've started ... well, you know :)

If you do it the other way round, that is with nothing
conducting until the gate voltage is reached, i think
thats much better.

Heres a small extract from something i found at:

analysis points out the one problem with the
circuit of Figure 2, namely, the voltage at
which crowbarring occurs is very dependent on
the characteristics of the SCR gate threshold:
the SCR is guaranteed to be off with 200mV on
its gate, but not guaranteed to be on with
less than 1.5V. The 1.3V spread means that
designing the SCR to trip on at a minimum of
3.0V means that it may not turn on until 4.3V.

It points out that SCR trigger thresholds can vary quite
a lot, more than i thought actually. I have always taken
it to be between one and two volts somewhere, and at very
little current.

What i can say is that the trigger is always the same on
the device, although it may vary to another device.

I use crowbar clampdown on my own home made supplies, and
have also used gate repeating caps on some. This is an
arrangement where the resistor is just a little too high
for 'hold-on' current within the supply, and the SCR drops
out of conduction, allowing the zener to re-instate the
stabilised voltage. On one such supply i made, i used a
small LS and it clicks like a buzzy-bee if i short the
supply into over-current until i remove the short and then
the voltage rises back to its stabilised setting.

The same article mentions 2 microseconds for the SCR to
operate, well that sounds awful slow to me, if its that
slow, then maybe the 2222 would actually have time to
respond to the 555 and go into conduction before the
SCRs have a chance to engage !
Which of course is what that circuit requires.

But i still prefer to have the SCRs just waiting for their
gates to come up, with no other bits active. The R-C work
on the gate is quite normal, as you say a second is a bit
longer than usual, thats why i would put a limiting res
on the gate, because if the cap has to be bigger than say
5 or 6 mfd, that might be a bit much for the gate, so i
would put a series res just so it wouldn't damage the gate.

The circuit you have drawn looks perfectly workable, so
long as Q1 doesn't get a brief 'lo' from the 555 just as
its starting up.

Thanks for the interest, i don't get to chat much about
electronics. No-one round here where i live has any
interest in it. Well not that i've seen anyway.

I'm going to reduce the resolution on my monitor now cos
i'm running 98se and i worry about crashes. This resolution
uses more RAM, and i only have 64 meg in there.

Cheers, John :)

(Oh, and thanks for the shorter line lengths!!)
Yea i know what you are saying about the fact that Q1 would
actually be switched on if the 555 didnt respond fast enough,
that is why C4 is present.
Im not saying that is a bad set-up or anything with the cap and resistor,
i use it all the time, even to trigger the 555 in this diagram, the problem
i see is that to get a full one second delay the time constant must be
quite large, if it was in fact that large the cap would be very slow to
drain, so if the computer was shut down the user would have to wait
quite some time to restart. If you put a bleed off resistor it would
lower the discharge time, but make it quite difficult to get the charge
time correct.
Hey guys, I decided to go with the Relay circuit you designed originally with the +5 and +12V connectors (this decision was made because the person at the electronics component store mentioned to me that the SCR's would work fine, but would also lower the voltage output by around 0.5~1.5V which would lead to instability problems with the VGA card). I put it all together carefully sauldering each connection and I turned it on using a fan as my power tester and a PSU I had laying around. The circuit almost instantly flips the relays into the NO position which cuts current from the +5V and +12V rails. However, it never goes off, so I must have made some kind of an error. My suspicion is maybe I misread a point on the diagram and saudered a connection in the wrong place. Please advise me on my suspicion as I am going to attach a copy of your diagram with a circle around the circuit in question.
The 2nd place where I see a 3 way connection between the 1N4001's also makes me wonder if maybe I misread something there too. Effectively, wherever there was a dot, I saudered a wire connection.

Hmm upon looking again at the diagram, it also occurs to me that those 1N4001's might have polarity? Is this the case?in my circuit, the 12V flows through the black side toware the grey side which is connected to ground? Is this also


  • Delay_160.gif
    13.3 KB · Views: 1,461
yea all those dots mean they are soldered, i think you figured out why its not working, i think you put the diodes in wrong, It seems like you have D1 in correctly, but D2 may be in backwards, that would do exactly as you say. i will alter the picture for you to show what the diode would look like, the diodes are to supress EMF as the magnetic field in the relay collapses, thats why they may appear to be in backwards. Lets hope this works or john1 is gonna kill me :lol:

also are you sure you put the transistor in correctly?


  • Delay_revised.gif
    12.9 KB · Views: 644
Hi CVNET1 and 'e',

The circuit looks ok, but i haven't built it,
those diodes are to quench the 'kick' that comes from a sudden
alteration of current through an inductance. older circuits
would use a small capacitor to absorb it, but diodes are more

You could imagine the inductance rather like a little flywheel,
it takes a bit of a push to get going, once the current is
flowing its easier to push, its impedance (sort of resistance)
is a bit higher, that condition is reached within a few m-secs.

But when you stop pushing, it tries to drag you along a bit.
Yes, that inductor will try to 'suck' some more current through
if you switch it off. Mechanical switches for inductive loads
have to carry a higher rating than the normal running current.
This effect can cause arcing and sparking at switch contacts.

Where the switching is done by a semiconductor, the effect still
occurs, but its inside the device. So diodes are placed across
the inductor so that they will conduct and effectively 'short'
any reverse voltage that occurs, to protect the switching device
from the 'kick' that the inductor releases.

I have noticed a difficulty with some people regarding the
direction of the voltage surge. Sometimes a person will say:
"if the inductor is trying to continue or sustain the
flow of current through itself, -- then why isnt the voltage surge
in the the same direction?"

Well, this shows that the person asking is actually thinking.
There are a couple of answers to this, to show that the surge volts
are actually in reverse to the applied voltage. Sometimes its better
to let the person think it out for themselves, sometimes its better
to try to explain.

If you're pushing a cart along, then you stop, the cart will pull
you a little bit.
The effect is a reversal at the point where you and the cart meet.

There are more involved explanations about collapsing fields, and
cutting lines of flux, but that is usually enough to illustrate
that the surge is a reverse voltage.

That is why the diode is there.

Of course i hope it works fine, although i dont intend to make one.
My approach is a little more 'minimalistic'.

I was surprised at the shop comment about the voltage drop ...
ive never looked, maybe there is something there ...
'e' do you think thats valid ?

Hope this works out OK for you,
Regards, John :)
Thats a good explaination of the purpose of the kickback diode.
I was thinking about the drop on the SCR myself after i read that,
and i would assume it does drop about that considering the 4 layers,
Since a standard diode has 2 layers and drops about .7, double that
seems reasonable. Acording to the datasheet for the SCRs i used in
that diagram it typically drops 1.23 @ 5 A, so that sounds reasonable.
yes, i checked one of mine.
I got 0.8 volt. (AVO 8 )
at about 200mA

I will check some more, but i doubt that
would induce any video instability.

John :)
john1 said:
yes, i checked one of mine.
I got 0.8 volt. (AVO 8 )
at about 200mA

I will check some more, but i doubt that
would induce any video instability.

John :)
yea i dont think the power supply would be exact enough
for the diff to be a bad problem, although if the card connects
the two is would have a potential difference so maybe it would
cause a problem... :?: i doubt it, but it might...
Not open for further replies.

New Articles From Microcontroller Tips