# Source for SMD relays

Status
Not open for further replies.

#### ljcox

##### Well-Known Member
I need relays with the following specifications:-
A coil voltage in the range 5 Volt to 48 Volt.

Contacts - either DPST or DPDT able to switch at least 60 Volt DC and a current rating of at least 500 mA.
I prefer SMD mounting but through the hole will suffice if SMD relays with the above specifications are not available.

Does any one know where I can purchase such relays? I need 80, so the price is critical.

I tried RS Components & Element 14, but they have little to offer.

I also did a search of Chinese companies but found their web sites frustrating. They did not have a sensible search engine. Besides, I have the feeling they don't want enquiries from the likes of me.

Any assistance will be appreciated.
Len

#### Pommie

##### Well-Known Member
Is the 60V DC contact rating absolutely necessary? If not, banggood have 30V DC rated relays at 50c.

Mike.

#### ljcox

##### Well-Known Member
Thanks Mike,
Yes, 60 V is essential as I need to switch up to 56 Volt.

I had no problem finding 30 Volt relays.
I also had no problem finding relays that could switch up to 220 V DC.
My main problem was to find relays with a suitable coil & SMD.
Len

#### Rich D.

##### Active Member
I think dougy83's idea is a very good one. They should be more reliable and smaller. I wish I would have thought of that but I didn't.
I would recommend looking for something more rugged though. If you require 500mA and 60 volts to be switched, these provide no margin of error.
The specs state that 60V and 500mA is the "Absolute maximum", which means you are at the very edge of part failure.
As stated in the spec itself, "Using continuously under heavy loads (e.g. the application of high temperature/current/voltage and the significant change in temperature, etc.) may cause this product to decrease in the reliability significantly even if the operating conditions (i.e. operating temperature/current/voltage, etc.) are within the absolute maximum ratings."

I don't know the application. If these relays are needed in a design that only supplies small pulses of high currents, you can find a lot more devices that are rated for non-continuous current peaks. There is no way getting around the voltage requirements though. While excess current needs time to heat and destroy, voltage can blow-out something in nanoseconds.

Unfortunately I don't know about any specific parts that are SMT. I do know you can get solid-state relays that can handle much more current and voltage, but they are not usually very cheap - expect $5 to$10 or so, so that may be a bill of $400...$800. Yikes!
Also I suggest taking a look at 2-channel devices. If you can find the right one, you might be able to buy 1/2 the amount for a price improvement.

#### alec_t

##### Well-Known Member
Unless you need galvanic isolation, why not use just a MOSFET, such as one of these ?

#### ljcox

##### Well-Known Member
I think dougy83's idea is a very good one. They should be more reliable and smaller. I wish I would have thought of that but I didn't.
I would recommend looking for something more rugged though. If you require 500mA and 60 volts to be switched, these provide no margin of error.
The specs state that 60V and 500mA is the "Absolute maximum", which means you are at the very edge of part failure.
As stated in the spec itself, "Using continuously under heavy loads (e.g. the application of high temperature/current/voltage and the significant change in temperature, etc.) may cause this product to decrease in the reliability significantly even if the operating conditions (i.e. operating temperature/current/voltage, etc.) are within the absolute maximum ratings."
Yes, I understand that. The nominal voltage my circuit is to operate at is about -52 Volt. But it may on occasions be as high as -56 Volt. So although dough83's suggestion is a good one, it would be "sailing close to the wind".
I don't know the application. If these relays are needed in a design that only supplies small pulses of high currents, you can find a lot more devices that are rated for non-continuous current peaks. There is no way getting around the voltage requirements though. While excess current needs time to heat and destroy, voltage can blow-out something in nanoseconds.

Unfortunately I don't know about any specific parts that are SMT. I do know you can get solid-state relays that can handle much more current and voltage, but they are not usually very cheap - expect $5 to$10 or so, so that may be a bill of $400...$800. Yikes!
Also I suggest taking a look at 2-channel devices. If you can find the right one, you might be able to buy 1/2 the amount for a price improvement.
Later I'll post some of the options I have been considering so you can see what I'm trying to do.

Thanks for the replies.
Len

#### ljcox

##### Well-Known Member
Thanks for all of the suggestions.
After much consideration of the various options, I decided to avoid relays & do the holding function in the PIC.
I had intended to use a pair of 16F684 PICs but this PIC does not have enough input/output pins.
So I will use the 40 pin 16F877. The software will be more complicated as it only has 4 RB change inputs.
Len

#### Pommie

##### Well-Known Member
Len, if you switch to a more recent chip such as the 16F15376 then most of the I/O pins have IOC. You also get more ram/rom and double the stack space plus lots of other goodies.

Mike.

#### ljcox

##### Well-Known Member
Thanks Mike,
The 16F14375 would be ideal as it is very similar to the 16F877 (40 pin DIP). And it has IOC on many pins - as you said.

But neither Element 14 nor RS Components stock the DIP version.. Digikey do but the shipping cost is \$24.

Do you know where I can buy them? If all else fails, I'll buy the SMD version. But I have done the PCB design assuming a 16F877 so there will be a bit of work to change it to a 44 pin.

Cheers, Len

#### Pommie

##### Well-Known Member
I normally just buy direct from Microchip. However, if you just need 1 or 2 that could work out expensive.

Mike.

#### jpanhalt

##### Well-Known Member
@Len,

Do you have a .edu e-mail or know someone who does? I am staunchly opposed to abusing the generosity of vendors, but getting a chip as a "sample" can reduce shipping and handling expenses. You might even try to get a sample without an edu account. That characterization still might reduce shipping and handling.

John

#### ljcox

##### Well-Known Member
Mike, Yes I only need 6 so it would be expensive. Also, I looked at the spec for the 16F15375 & it looked daunting - something like 580 pages.

So I'll use the 16F877 & use an OR gate to do the IOC using one of the RB 7:4 inputs which have IOC.

John,
Thanks for the hint, but I don't have a .edu email. As you will see in the above, I've decided to use the 16F877.
Len

#### Pommie

##### Well-Known Member
Len, depending on how fast the inputs change, you could setup a timer interrupt every (say) 5mS and do the interrupt on change bit in code. I've done this in the past and not found a mechanical system that's too fast.
The code is also fairly simple to implement.

Mike.

#### ljcox

##### Well-Known Member
Thanks Mike, that's essentially what I was planning to do until yesterday when I had a brain wave.
It occurred to me that I could alter the hardware in order to detect a change & feed a signal into RB7.
As you know, RB7:4 have IOC. It requires a transistor, diode & a resistor.
The reason I want to use IOC is so I can put the PIC to sleep when not in use.
However, today (partly as a result of your post) I'm wondering whether the extra hardware is worth it. I assume the PIC will be happy to keep churning away 24 Hr/Day doing virtually nothing.
What do you think?

I'm designing a device for the Telstra Museum. It is only open on the last Wednesday of each month so the device will only be used on those days.
The Telstra Museum is well worth seeing for anyone who has an interest in the history of telephony technology. They have almost the full progression of technology from the 1880's to about 1995.
Most of it is in working order so the volunteers who are there can demonstrate the various items to visitors.
There are some gaps but it is well worth a look. It is located in the Hawthorn Exchange building in Burwood Rd about 100 metre from the Glenferrie Rd intersection. It is directly south of the Glenferrie station so access by train is easy (about 3 minutes walk).

I believe there is a Telstra Museum in Brisbane. I don't know anything about it. Have you been there?
Len

#### Pommie

##### Well-Known Member
I did a battery powered project that slept most of the time. It was awakened by timer1 every 5mS, checked if anything needed doing and promply went back to sleep. The timer was set to use the 32k internal clock - however, I don't think that is available on the earlier chips. At school so can't check.

To do edge detection in software requires keeping a copy of the ports previous state and then xoring with the current state gives you pins that have changed. Anding the changed pins with the current value gives rising edges. Anding the changed pins with the previous value gives falling edges.

I'm not aware on the Telstra Museum in Brisbane but will certainly look it up.

Mike.

#### ljcox

##### Well-Known Member
Thanks Mike, I had not thought of using a timer to wake the PIC.

Yes, I know how to do the edge detection thanks. I have had a prototype working for several weeks. But it uses the `16F684 & could cope with only 10 P wires.
Thus I would need to produce 7 PCBs. (They want to serve a total of 72 P wires)
Using the 16F877 means I can put 12 P wires on each PCB thus reducing the total to 6. I can also use the "holding" function rather than have reset inputs.

Len

#### ljcox

##### Well-Known Member
I did a battery powered project that slept most of the time. It was awakened by timer1 every 5mS, checked if anything needed doing and promply went back to sleep. The timer was set to use the 32k internal clock - however, I don't think that is available on the earlier chips. At school so can't check.

Mike.
Mike, I looked at the spec for the 16F877 but the only way I can see to wake from sleep using the TMR1 overflow interrupt is to have TMR1 driven by an external clock.
Or have I missed something?
Len