Can anyone point me to a clear, well written guide to relay
specifications, particularly with regard to reactive loads?
I'm looking at a relay in a design, and it's not clear whether or not
it is being used within spec. Inrush current is ok, but peak voltage
is high (few uS at opening).
One source tells me that because Ipeak * Vpeak is larger than the kVA
rating that there is a problem, but it doesn't make sense to me in
that Ipeak and Vpeak cannot occur at the same time. My Ipeak and
Vpeak measurements are under absolute worst case conditions, measuring
many times with the scope, and taking the worst observed value.
The more I dig, the fuzzier the ratings seem to become.
If you find such a thorough explanation, please let me know :)
We (coworkers and I) had an issue where we had to find a contactor
which could switch 60VDC and 80 Amps peak. Using the ratings at face
value, we were going to have to use a contactor which was the size of
a lawnmower engine (and no doubt cost several times as much). However,
after working with the applications engineers at the company (I think
it was Rockwell Automation/Allen Bradley), they were able to tell us
that we could put several contacts in series and effectively raise the
interrupting DC voltage and current specs of the contacts by doing so.
We provided them information on how the 80 Amps current was only in
0.5 second pulses and in-between the current was no more than 10 Amps.
I do not remember whether they took that into account or not.
I have since seen a few contactor catalogs which give tables showing
the ratings when you have 2,3,4 etc. contacts in series, but certainly
no great discussion of the whys and hows.
We didn't have a reactive load but most contactors have not only
voltage and current ratings but also a maximum L/R ratio for the
circuit. If you have a good handle on what the inductance is, you
should be able to see if this rating is being exceeded.
Is this an AC or DC application?
The whole Ipk * Vpk > kVA thing sounds bogus to me. Sounds like you
need to find a different engineer to talk to at that vendor.
> Can anyone point me to a clear, well written guide to relay
> specifications, particularly with regard to reactive loads?
>
> I'm looking at a relay in a design, and it's not clear whether or not
> it is being used within spec. Inrush current is ok, but peak voltage
> is high (few uS at opening).
> One source tells me that because Ipeak * Vpeak is larger than the kVA
> rating that there is a problem, but it doesn't make sense to me in
> that Ipeak and Vpeak cannot occur at the same time. My Ipeak and
> Vpeak measurements are under absolute worst case conditions, measuring
> many times with the scope, and taking the worst observed value.
>
> The more I dig, the fuzzier the ratings seem to become..
On Tue, Dec 21, 2010 at 1:07 PM, David VanHorn <.....microbrixKILLspam@spam@gmail.com> wrote:
> Can anyone point me to a clear, well written guide to relay
> specifications, particularly with regard to reactive loads?
>
> I'm looking at a relay in a design, and it's not clear whether or not
> it is being used within spec. Inrush current is ok, but peak voltage
> is high (few uS at opening).
> One source tells me that because Ipeak * Vpeak is larger than the kVA
> rating that there is a problem, but it doesn't make sense to me in
> that Ipeak and Vpeak cannot occur at the same time. My Ipeak and
> Vpeak measurements are under absolute worst case conditions, measuring
> many times with the scope, and taking the worst observed value.
>
> The more I dig, the fuzzier the ratings seem to become..
Part of the problem is that relays wear out.
If the relay is overloaded, it will not last as long, but relays don't
last forever. If it is slightly overloaded, the life will be reduced
slightly. Even grossly overloaded, the relay might work for a few
cycles before the contacts burn out. So you are moving along a curve,
not going from a binary "in spec, works" to "out of spec, fails."
On 12/21/2010 1:22 PM, Sean Breheny wrote: {Quote hidden}
> Hi Dave,
>
> If you find such a thorough explanation, please let me know :)
>
> We (coworkers and I) had an issue where we had to find a contactor
> which could switch 60VDC and 80 Amps peak. Using the ratings at face
> value, we were going to have to use a contactor which was the size of
> a lawnmower engine (and no doubt cost several times as much). However,
> after working with the applications engineers at the company (I think
> it was Rockwell Automation/Allen Bradley), they were able to tell us
> that we could put several contacts in series and effectively raise the
> interrupting DC voltage and current specs of the contacts by doing so.
> We provided them information on how the 80 Amps current was only in
> 0.5 second pulses and in-between the current was no more than 10 Amps.
> I do not remember whether they took that into account or not.
>
> I have since seen a few contactor catalogs which give tables showing
> the ratings when you have 2,3,4 etc. contacts in series, but certainly
> no great discussion of the whys and hows.
>
> We didn't have a reactive load but most contactors have not only
> voltage and current ratings but also a maximum L/R ratio for the
> circuit. If you have a good handle on what the inductance is, you
> should be able to see if this rating is being exceeded.
>
> Is this an AC or DC application?
>
> The whole Ipk * Vpk> kVA thing sounds bogus to me. Sounds like you
> need to find a different engineer to talk to at that vendor.
>
> Sean
>
>
> On Tue, Dec 21, 2010 at 2:07 PM, David VanHorn<EraseMEmicrobrixspam_OUTTakeThisOuTgmail.com> wrote:
>> Can anyone point me to a clear, well written guide to relay
>> specifications, particularly with regard to reactive loads?
>>
>> I'm looking at a relay in a design, and it's not clear whether or not
>> it is being used within spec. Inrush current is ok, but peak voltage
>> is high (few uS at opening).
>> One source tells me that because Ipeak * Vpeak is larger than the kVA
>> rating that there is a problem, but it doesn't make sense to me in
>> that Ipeak and Vpeak cannot occur at the same time. My Ipeak and
>> Vpeak measurements are under absolute worst case conditions, measuring
>> many times with the scope, and taking the worst observed value.
>>
>> The more I dig, the fuzzier the ratings seem to become..
> If the relay is overloaded, it will not last as long, but relays don't
> last forever. If it is slightly overloaded, the life will be reduced
> slightly. Even grossly overloaded, the relay might work for a few
> cycles before the contacts burn out. So you are moving along a curve,
> not going from a binary "in spec, works" to "out of spec, fails."
Agreed, but I need a more deterministic path than "Try it and see what happens".
For this application, 10k cycles under full load would be a lifetime.
> The whole Ipk * Vpk > kVA thing sounds bogus to me. Sounds like you
> need to find a different engineer to talk to at that vendor.
According to them, this is THE guy, 30+ years.
They strongly recommend NOT using the standard parallel diode clamp
that everyone uses, and I can see why, it causes the release time to
be significantly slower. It's hard to see on a scope because you have
to measure from coil drive stop to when the relay begins to release,
but I can see it, and I'm getting about half the time with a zener
snubber than with the parallel diode.
That doesn't address my contact rating issue, but it does help contact
life since the contacts don't spend as much time arcing
> Can anyone point me to a clear, well written guide to relay
> specifications, particularly with regard to reactive loads?
>
> I'm looking at a relay in a design, and it's not clear whether or not
> it is being used within spec. Inrush current is ok, but peak voltage
> is high (few uS at opening).
> One source tells me that because Ipeak * Vpeak is larger than the kVA
> rating that there is a problem, but it doesn't make sense to me in
> that Ipeak and Vpeak cannot occur at the same time. My Ipeak and
> Vpeak measurements are under absolute worst case conditions, measuring
> many times with the scope, and taking the worst observed value.
>
> The more I dig, the fuzzier the ratings seem to become..
Not an exact pointer, but there is some good reading on relays in these Tyco application notes:
>They strongly recommend NOT using the standard parallel diode clamp
>that everyone uses, and I can see why, it causes the release time to
>be significantly slower. It's hard to see on a scope because you have
>to measure from coil drive stop to when the relay begins to release,
>but I can see it, and I'm getting about half the time with a zener
>snubber than with the parallel diode.
I'm fond of driving relays with TPIC6595 or TPIC6c595 octal shift registers /w built-in MOSFET drivers. Most of my boards use them for relay populations >4 or so.
They have built-in avalanche clamps set to about 60V. You do NOT need to use back-EMF diodes with the relays being driven by these devices.
As you note, the relays release MUCH faster when the coil voltage is allowed to spike way high rather than being clamped by a forward-conducting diode.
As an aside, I've been known to use the standard back-EMF clamp diode on relay coils when I *do* want them to hold in a little bit longer than normal. Those are mostly on systems where the relay PSU rail is chopped or has significant ripple. It may be cheating but its extremely reliable <grin>.
Ive seen the ratings for series contacts in the Rockwell automation
contactor catalog, they are talking about in a multi-pole contactor
(I.E, one meant for 3 phase AC with all the contacts put in series for
whatever DC load) And the particular ones I was looking at was
"safety" contactors where all the contacts were positively linked,
along with aux contacts for feedback, so the opening and closing of
those should be as coordinated as you'll get with a mechanical switch.
I'd imagine different drop-out speeds could indeed be an issue with
separate 1-pole contactors.
-Jon
> Did the switch timing differences between the contactors have any
> bearing on the outcome?
>
> Mark Skeels
> Engineer
> Competition Electronics, Inc.
> TEL: 815-874-8001
> FAX: 815-874-8181
> http://www.competitionelectronics.com
>
> On 12/21/2010 1:22 PM, Sean Breheny wrote:
>> Hi Dave,
>>
>> If you find such a thorough explanation, please let me know :)
>>
>> We (coworkers and I) had an issue where we had to find a contactor
>> which could switch 60VDC and 80 Amps peak. Using the ratings at face
>> value, we were going to have to use a contactor which was the size of
>> a lawnmower engine (and no doubt cost several times as much). However,
>> after working with the applications engineers at the company (I think
>> it was Rockwell Automation/Allen Bradley), they were able to tell us
>> that we could put several contacts in series and effectively raise the
>> interrupting DC voltage and current specs of the contacts by doing so.
>> Â We provided them information on how the 80 Amps current was only in
>> 0.5 second pulses and in-between the current was no more than 10 Amps.
>> I do not remember whether they took that into account or not.
>>
>> I have since seen a few contactor catalogs which give tables showing
>> the ratings when you have 2,3,4 etc. contacts in series, but certainly
>> no great discussion of the whys and hows.
>>
>> We didn't have a reactive load but most contactors have not only
>> voltage and current ratings but also a maximum L/R ratio for the
>> circuit. If you have a good handle on what the inductance is, you
>> should be able to see if this rating is being exceeded.
>>
>> Is this an AC or DC application?
>>
>> The whole Ipk * Vpk> Â kVA thing sounds bogus to me. Sounds like you
>> need to find a different engineer to talk to at that vendor.
>>
>> Sean
>>
>>
>> On Tue, Dec 21, 2010 at 2:07 PM, David VanHorn<RemoveMEmicrobrixTakeThisOuTgmail.com> Â wrote:
>>> Can anyone point me to a clear, well written guide to relay
>>> specifications, particularly with regard to reactive loads?
>>>
>>> I'm looking at a relay in a design, and it's not clear whether or not
>>> it is being used within spec. Â Inrush current is ok, but peak voltage
>>> is high (few uS at opening).
>>> One source tells me that because Ipeak * Vpeak is larger than the kVA
>>> rating that there is a problem, but it doesn't make sense to me in
>>> that Ipeak and Vpeak cannot occur at the same time. Â My Ipeak and
>>> Vpeak measurements are under absolute worst case conditions, measuring
>>> many times with the scope, and taking the worst observed value.
>>>
>>> The more I dig, the fuzzier the ratings seem to become..
> They strongly recommend NOT using the standard parallel
> diode clamp
> half the time with a zener snubber than with the parallel diode
I saw a schematic recently where someone was using a (1N4001
back-to-back with a 20V zener) in parallel with the coil. He
claimed that this caused the relay to drop out at 2ms rather than
9ms. It was a ham radio circuit, one suggestion being that faster
drop-out could be useful when keying Morse for example
> They strongly recommend NOT using the standard parallel diode clamp
> that everyone uses, and I can see why, it causes the release time to
> be significantly slower. It's hard to see on a scope because you have
> to measure from coil drive stop to when the relay begins to release,
> but I can see it, and I'm getting about half the time with a zener
> snubber than with the parallel diode.
They are answering the wrong question if the problem is with the delay per se.
This is a design issue and release time can be tolerated or adjusted
as required.
IF they were commenting on release time because it impacted the
contact rating issue - and they MAY be - then it would be a legitimate
comment.
Release time has a few second order effects, but the first order
aspect is easily understood and designed for. A relay has a "hold in
voltage" which is a proxy for a hold in current which is a proxy for a
hold in field strength which is ... We'll stop at the hold in current
point.
When the relay drive ceases (assume a sudden "square" drive stop) the
current in the coil just after turnoff would be EXACTLY the same as
the current just before turnoff for a perfectly inductive-only coil.
In the real world the current will step slightly instantaneously due
to capacitance and other effects. But close enough Ion = Ioff.
Irelease is (usually) known. The relay will release when Irelay falls
to Irelease.
Current will decay with energy loss. Know the rate of energy looss and
you can predict current change and thus voltage change and thus time
to dropout.
With a clamp diode the loss is mainly Vdiode x Irelay + Irelay^2 x
Rrelay. The latter term may or may not be significant.
With a zener the loss is Vzener x Irelay Irelay^2 x Rrelay.
If Vzener >> Vdiode then time to dissipate enough energy to achieve
dropout with the zener is going to be shorter to much shorter.
Note that this effect is liable to be more pronounced with an AC relay
than a DC relay because the AC relay is designed to drop a
significant proportion of it's operating voltage across the
inductance, so the resistance per rated current is lower than for a DC
relay. (This doesn't alter the current being carried, just the
relative losses).
AC relays, of course, tend to be unhappy about diodes connected across
their coils, which may relate to Joe's back to back diode post.
An alternative to diode or zener is a resistor across the coil. The
current will flow trough the resistor and dissipate energy. Rate of
dissipation is related to initial I^@R and voltage rise at turn off to
IR.
More could be written ... :-.
> That doesn't address my contact rating issue, but it does help contact
> life since the contacts don't spend as much time arcing.
As above - that MAY be why they said it. But, relay will tend to hold
in until dropout voltage is reached and then will dropout relatively
rapidly regardless. Relatively.
> > They strongly recommend NOT using the standard parallel
> > diode clamp
> > half the time with a zener snubber than with the parallel diode
>
> I saw a schematic recently where someone was using a (1N4001
> back-to-back with a 20V zener) in parallel with the coil. He
> claimed that this caused the relay to drop out at 2ms rather than
> 9ms. It was a ham radio circuit, one suggestion being that faster
> drop-out could be useful when keying Morse for example
I've just been playing with something similar this week. Pneumatic solenoid valve 12V DC 0.5A (electrically just like a relay), and in this case I wanted a very fast turn off. Using a smart high side switch, Infineon ITS/BTS4140N, to drive the coil. These devices have a negative voltage clamping mechanism for inductive loads which operates at around -62 to -68V, so no clamping diode is necessary.
With a scope I see at turn off the voltage accross the coil go from +12V to -64V almost instantly, clamp at that for about 0.25ms then decay towards zero. Very good, just what I wanted.
To satisfy my own curiosity I added a 36V AC TVS diode across the coil. Turn off waveform is as above, but -ve peak is now approx 36V and clamps for approx 0.5ms before decaying. So, half the peak voltage but 2 x slower. Also good, fast enough for what I need and now the TVS diode is absorbing the turn off energy not the switching device.
Had I known this thread was coming up I would have also compared with a 1N4007 diode across the coil. I would expect the turn off voltage to clamp at approx -0.7V for maybe some tens of ms... too slow.
Getting the solenoid valve to stop flowing liquid in a few ms or even tens of ms is another story... still working on that one~!
-- Brent Brown, Electronic Design Solutions
16 English Street, St Andrews,
Hamilton 3200, New Zealand
Ph: +64 7 849 0069
Fax: +64 7 849 0071
Cell: +64 27 433 4069
eMail: spamBeGonebrent.brownspamBeGoneclear.net.nz
Much of what I said, then, is not applicable to this because I was
assuming a DC application. If the voltage and current though the
contacts has zero average (i.e., no DC component), then you should be
able to use the AC specs which tend to be much more "fleshed out" in
datasheets and catalogs than the DC ones.
You mentioned a microsecond or so long voltage spike. You are talking
about a spike across the contacts, right? Did you measure this?
My understanding for AC contactor ratings is as follows: You first
determine the "class" of operation you need (e.g., a purely resistive
load might be class I, inductive load with L/R < 1 millisecond rarely
switched under load might be class II, a highly inductive load which
is switched on and off constantly, like a motor being controlled by a
contactor, might be class III). I don't remember the exact terms here
but they are usually spelled out in application notes or the catalog
page. Once you have a class designation, you look at the voltage,
current, and VA ratings for your contactor when used in this class.
You then take the maximum voltage and current which you will apply to
the relay quasi-statically (i.e., you should not need to consider
spikes on opening since they are already rolled into the rating for
the class) and make sure that voltage<Vmax, current<Imax, and
voltage*current<VAmax.
There would also be a consideration for inrush current if there is
capacitance or other circuit elements which might allow inrush current
(but you said that you already have taken that into account).
>> Is this an AC or DC application?
>
> AC
>
>> The whole Ipk * Vpk > kVA thing sounds bogus to me. Sounds like you
>> need to find a different engineer to talk to at that vendor.
>
> According to them, this is THE guy, 30+ years.
>
> They strongly recommend NOT using the standard parallel diode clamp
> that everyone uses, and I can see why, it causes the release time to
> be significantly slower. It's hard to see on a scope because you have
> to measure from coil drive stop to when the relay begins to release,
> but I can see it, and I'm getting about half the time with a zener
> snubber than with the parallel diode.
>
> That doesn't address my contact rating issue, but it does help contact
> life since the contacts don't spend as much time arcing.
My recent experience was with safety contactors so Jonathan is right.
I do not know whether the contacts-in-series trick can be done with
other types. However, I think that the mechanism of action here
includes arcing. In other words, placing several contacts in series
doesn't eliminate arcing, it merely reduces the maximum duration of
the arc to the time it takes for all of the contacts to open. When the
first one opens, an arc will start and there will be a voltage drop
associated with it. When the second one opens, there may then not be
sufficient remaining voltage to sustain an arc across BOTH sets of
contacts and if so, the arc will cease.
> Ive seen the ratings for series contacts in the Rockwell automation
> contactor catalog, they are talking about in a multi-pole contactor
> (I.E, one meant for 3 phase AC with all the contacts put in series for
> whatever DC load) And the particular ones I was looking at was
> "safety" contactors where all the contacts were positively linked,
> along with aux contacts for feedback, so the opening and closing of
> those should be as coordinated as you'll get with a mechanical switch.
> I'd imagine different drop-out speeds could indeed be an issue with
> separate 1-pole contactors.
> -Jon
>
> On Tue, Dec 21, 2010 at 3:43 PM, Mark E. Skeels
> <mskeelsEraseME.....competitionelectronics.com> wrote:
>> Did the switch timing differences between the contactors have any
>> bearing on the outcome?
>>
>> Mark Skeels
>> Engineer
>> Competition Electronics, Inc.
>> TEL: 815-874-8001
>> FAX: 815-874-8181
>> http://www.competitionelectronics.com
>>
>> On 12/21/2010 1:22 PM, Sean Breheny wrote:
>>> Hi Dave,
>>>
>>> If you find such a thorough explanation, please let me know :)
>>>
>>> We (coworkers and I) had an issue where we had to find a contactor
>>> which could switch 60VDC and 80 Amps peak. Using the ratings at face
>>> value, we were going to have to use a contactor which was the size of
>>> a lawnmower engine (and no doubt cost several times as much). However,
>>> after working with the applications engineers at the company (I think
>>> it was Rockwell Automation/Allen Bradley), they were able to tell us
>>> that we could put several contacts in series and effectively raise the
>>> interrupting DC voltage and current specs of the contacts by doing so.
>>> We provided them information on how the 80 Amps current was only in
>>> 0.5 second pulses and in-between the current was no more than 10 Amps.
>>> I do not remember whether they took that into account or not.
>>>
>>> I have since seen a few contactor catalogs which give tables showing
>>> the ratings when you have 2,3,4 etc. contacts in series, but certainly
>>> no great discussion of the whys and hows.
>>>
>>> We didn't have a reactive load but most contactors have not only
>>> voltage and current ratings but also a maximum L/R ratio for the
>>> circuit. If you have a good handle on what the inductance is, you
>>> should be able to see if this rating is being exceeded.
>>>
>>> Is this an AC or DC application?
>>>
>>> The whole Ipk * Vpk> kVA thing sounds bogus to me. Sounds like you
>>> need to find a different engineer to talk to at that vendor.
>>>
>>> Sean
>>>
>>>
>>> On Tue, Dec 21, 2010 at 2:07 PM, David VanHorn<EraseMEmicrobrixgmail.com> wrote:
>>>> Can anyone point me to a clear, well written guide to relay
>>>> specifications, particularly with regard to reactive loads?
>>>>
>>>> I'm looking at a relay in a design, and it's not clear whether or not
>>>> it is being used within spec. Inrush current is ok, but peak voltage
>>>> is high (few uS at opening).
>>>> One source tells me that because Ipeak * Vpeak is larger than the kVA
>>>> rating that there is a problem, but it doesn't make sense to me in
>>>> that Ipeak and Vpeak cannot occur at the same time. My Ipeak and
>>>> Vpeak measurements are under absolute worst case conditions, measuring
>>>> many times with the scope, and taking the worst observed value.
>>>>
>>>> The more I dig, the fuzzier the ratings seem to become..
>>>> --
On 22 December 2010 16:07, Sean Breheny <RemoveMEshb7EraseMEEraseMEcornell.edu> wrote:
> My recent experience was with safety contactors so Jonathan is right.
> I do not know whether the contacts-in-series trick can be done with
> other types. However, I think that the mechanism of action here
> includes arcing. In other words, placing several contacts in series
> doesn't eliminate arcing, it merely reduces the maximum duration of
> the arc to the time it takes for all of the contacts to open. When the
> first one opens, an arc will start and there will be a voltage drop
> associated with it. When the second one opens, there may then not be
> sufficient remaining voltage to sustain an arc across BOTH sets of
> contacts and if so, the arc will cease.
>
> Sean
>
One other effect worth noting is the difference between a relay and a
contactor. With a contactor designed for reasonable current, the
mechanical operation is spring loaded with a bit of play deliberately
built in. This way, on disconnect the armature has got a fair amount
of speed up before it impacts the contact carrier, pushing the
contacts apart at a higher rate to minimise any arc.
And in some cases it may be desirable to delay the disconnect.
Fo example in case of a fault condition, it's generally better to make
sure the fuse or breaker has opened before the contactor disconnects.
Otherwise the fault current can create a significant arc.
One other effect of high fault currents can be that the magnetic field
that develops can tend to force the contacts open. So the coil current
needs to be sufficient to withstand this additional force.
High current DC contactors can also have magnetic blowouts also, which
blow the arc outwards, increase its length and keeping it away from
the metalwork. It only works if the current is only passing in one
direction - not recommended for battery connections.
> > If the relay is overloaded, it will not last as long, but relays don't
> > last forever. If it is slightly overloaded, the life will be reduced
> > slightly. Even grossly overloaded, the relay might work for a few
> > cycles before the contacts burn out. So you are moving along a curve,
> > not going from a binary "in spec, works" to "out of spec, fails."
>
>
> Agreed, but I need a more deterministic path than "Try it and see what happens".
>
> For this application, 10k cycles under full load would be a lifetime.
A common way of dealing with a high current application is to have two sets of contacts in parallel, and set up so one contact closes before the other, and the one that closes late opens early on release. This then allows the late close/early open contacts to do their thing without arcing, and so carry most of the current in normal operation, while the early close/late open contacts become sacrificial in the closing current peak, and opening arc. I believe a properly set up relay uses different contact materials for the two sets, which are designed for the respective purposes.
-- Scanned by iCritical.
RussellMc wrote:
> They are answering the wrong question if the problem is with the
> delay per se. This is a design issue and release time can be
> tolerated or adjusted as required.
I didn't take it as meaning the big issue was the delay from telling the
relay to open until it opens, but rather the speed of opening. Once the
relay starts to open, you want it to open completely as fast as possible.
This is to transitions thru the potentially damaging arc region quickly.
> An alternative to diode or zener is a resistor across the coil. The
> current will flow trough the resistor and dissipate energy.
Usually a diode is added in series with the resistor so that no extra
current is drawn when the relay is being held closed.
Other tricks I've seen is to drop the coil current to just at the holding
current shortly before you want to open the relay. You usually want
somewhat more than the minimum holding current when held closed unless
you're sure you don't have vibration in your environment.
Another trick is to put a SCR in parallel with the relay contacts. Turn on
the SCR during the last half cycle of the AC you are trying to switch and
open the relay at the same time. The relay opens some time during the half
cycle, but the SCR takes over conduction until the current zero crossing.
HP made a power switching unit based on this principal that had incredibly
long relay life because the contacts were always opening with a SCR accross
them taking the current, so no arcing. The reverse was done when closing
the relay.
>> An alternative to diode or zener is a resistor across the coil. The
>> current will flow trough the resistor and dissipate energy.
> Usually a diode is added in series with the resistor so that no extra
> current is drawn when the relay is being held closed.
We looked at this, but we also have to figure in the $0.05/placement
cost, and that makes us favor a dual zener package to snub both relay
coils. Even with a dual resistor pack, we'd be looking at two
placements. Frequently for us, placement costs far exceed the
component costs. Space is also an issue, but not so critical. We are
also constrained by the total power supply energy available, we can't
afford the additional power to add resistors alone without the series
diodes across the coil, because in order to do any real good they
would have to be low enough that we can't afford the extra current at
low line voltage.
Constraints make engineering fun. :-P
> Other tricks I've seen is to drop the coil current to just at the holding
> current shortly before you want to open the relay. You usually want
> somewhat more than the minimum holding current when held closed unless
> you're sure you don't have vibration in your environment.
Good one to keep up your sleeve, and we thought about it, but rejected
because of additional component placement costs.
> Another trick is to put a SCR in parallel with the relay contacts. Turn on
> the SCR during the last half cycle of the AC you are trying to switch and
> open the relay at the same time. The relay opens some time during the half
> cycle, but the SCR takes over conduction until the current zero crossing.
> HP made a power switching unit based on this principal that had incredibly
> long relay life because the contacts were always opening with a SCR accross
> them taking the current, so no arcing. The reverse was done when closing
> the relay.
We will get close to this by looking at the AC voltage waveform, and
adjusting the contact opening time such that it happens when the
current is approaching minimum. If we're a tad late, then it only gets
better.
David VanHorn wrote:
>> Usually a diode is added in series with the resistor so that no extra
>> current is drawn when the relay is being held closed.
>
> We looked at this, but we also have to figure in the $0.05/placement
> cost, and that makes us favor a dual zener package to snub both relay
> coils.
Is 5 cents really relevant compared to the cost of a relay? And 5 cents is
way out of line for placement cost of a single resistor.
> Frequently for us, placement costs far exceed the
> component costs.
You've got the tail wagging the dog. It sounds like you should be looking
for a better manufacturer. 5 cents placement cost for a simple 0805 part is
out of line.
> Is 5 cents really relevant compared to the cost of a relay?
Our relays are about $0.25 plus placement.
>> Frequently for us, placement costs far exceed the
>> component costs.
>
> You've got the tail wagging the dog. It sounds like you should be looking
> for a better manufacturer. 5 cents placement cost for a simple 0805 part is
> out of line.
That is the official figure to be used in budgeting.
"This has been willed where what is willed must be.." (bonus points
if you know the quote.)
> -------- Original Message --------
> Subject: Re: [EE] Relay specifications
> From: David VanHorn <RemoveMEmicrobrixspam_OUTKILLspamgmail.com>
> Date: Tue, December 28, 2010 2:24 pm
> To: "Microcontroller discussion list - Public." <RemoveMEpiclistTakeThisOuTspammit.edu>
>
>
> > Is 5 cents really relevant compared to the cost of a relay?
>
> Our relays are about $0.25 plus placement.
>
> >> Frequently for us, placement costs far exceed the
> >> component costs.
> >
> > You've got the tail wagging the dog. Â It sounds like you should be looking
> > for a better manufacturer. Â 5 cents placement cost for a simple 0805 part is
> > out of line.
>
> That is the official figure to be used in budgeting.
> "This has been willed where what is willed must be.." (bonus points
> if you know the quote.)
>
> > We looked at this, but we also have to figure in the $0.05/placement cost,
> 5 cents placement cost for a simple 0805 part is out of line.
Is that what was actually meant?
It would be high in many cases.
One in-China manufacturer charges us 1/3 cent per "pin" - a somewhat
arbitrary swings-and roundabouts measure that nonetheless dominates
the cost of resistors and almost doubles the cost of small
transistors. (Resistors cost << 1 cent and installation costs 2/3
cent. BC807 and similar cost about 1.2 cents-ish and installation
costs 1 cent. A 14 pin SOIC is a disaster :-).
At 01:51 AM 12/29/2010, you wrote:
> > > We looked at this, but we also have to figure in the
> $0.05/placement cost,
>
> > 5 cents placement cost for a simple 0805 part is out of line.
>
>Is that what was actually meant?
>It would be high in many cases.
This is one of those "it depends" things. $0.50-$1 a part for
small/prototype quantities (say 3-5 pieces of boards with 300 parts) is not
necessarily out of line. Inclusive of setup, fast turnaround, pickup/drop off
etc. At the other end, the _variable_ cost of mounting one more 0603 part on an
automated line in large volume is probably << 0.1 cent (despite arbitrary
charges from assemblers).
gmail.com> wrote:
