I'm trying to hang an 8-way relay-driver off a USB port, with the circuit being bus-powered. 

The relays in question need at least a 9v signal, and they can be at the end of a 25' wire (DB9<-~~->DP9). This part is fixed because it's an off-the shelf part.

The first attempt was to try using an ICL7660 (since I had them lying around) in voltage-doubling configuration. That worked fine until I actually wanted to use the relays [grin]. The 7660 could produce 9v with no load, but as soon as I connected just one of the relay pins, the voltage dropped to ~4v and the relay failed to switch.

So I was wondering if the MAX232 would do any better - I know it's not what it's supposed to be used for, but all I really want is +10v from the USB's +5v that will drive 8 relays... I'm supposing that the RS232 drivers would probably be specced to drive the voltage down a long serial wire anyway, so maybe they'll do better ?

If it comes down to it, I can have a separate power-input to the controller (and basically make it not bus-powered) and then there's no problem, but if there's a good way of avoiding that I'd like to know about it :)

Thoughts, anyone ?

Cheers
   Simon

Simon wrote:
> I'm trying to hang an 8-way relay-driver off a USB port, with the
> circuit being bus-powered.
>
> The relays in question need at least a 9v signal, and they can be at
> the end of a 25' wire (DB9<-~~->DP9). This part is fixed because it's
> an off-the shelf part.
>
> The first attempt was to try using an ICL7660 (since I had them lying
> around) in voltage-doubling configuration. That worked fine until I
> actually wanted to use the relays [grin]. The 7660 could produce 9v
> with no load, but as soon as I connected just one of the relay pins,
> the voltage dropped to ~4v and the relay failed to switch.
>

Wait - that's a *current* problem, not a *voltage* problem. the
effective impedance/resistance of the thing is too low to drive
with that source. It's really a 4V source with that load; it's just
lying through your voltmeter when you test it no-load. That's okay;
they all do that.

> So I was wondering if the MAX232 would do any better - I know it's
> not what it's supposed to be used for, but all I really want is +10v
> from the USB's +5v that will drive 8 relays... I'm supposing that the
> RS232 drivers would probably be specced to drive the voltage down a
> long serial wire anyway, so maybe they'll do better ?
>

It might - it has a builtin charge pump. I would not trust one past 5V.

But can't you add another relay at the PC end
with enough energy? That would save you having a wallwart at the
far end.

> If it comes down to it, I can have a separate power-input to the
> controller (and basically make it not bus-powered) and then there's
> no problem, but if there's a good way of avoiding that I'd like to
> know about it :)
>
> Thoughts, anyone ?
>
> Cheers Simon
>

--
Les Cargill

On Monday, August 27, 2012 4:37:05 PM UTC-7, Les Cargill wrote:
> 
> It might - it has a builtin charge pump. I would not trust one past 5V.

Hmm - not past 5v, huh...

Another option I thought of would be something like ( http://www.digikey.com/product-detail/en/VBT2-S5-S12-SMT/102-2078-1-ND/2295136 ) - although I'd have to put a 680R[*] resistor across the outputs to make sure there's always a 10% load on them even if no relays are switched on. Wasting power like that just seems .... wrong :) 

I'd assume that this power would be sufficient: The bench PSU reports I'm pulling 0.1A @ 5V into the '7660 circuit when I have all 8 relays switched on, and the above part says its max output is 0.167A. That ought to be enough headroom.

[*] working it out as requiring at least 17mA of current at 12v, so

                12 = (17/1000) * R
            => R = 705 = ~680
            and Power dissipated = ~0.2W 


> 
> But can't you add another relay at the PC end
> with enough energy? That would save you having a wallwart at the
> far end.

Yeah, the above isn't really as clear as I ought to have been. When I said "controller", I meant the USB interface, not the relay. The setup looks something like:

[embedded] <--USB--> [relay controller board] <--25' cable --> [mains-powered relays]
[  system   ]

Where the embedded system only has a couple of USB port outputs, and the relay controller board is (currently) powered from one of those outputs. It's an AVR coupled with an FT232 and some transistors to allow the AVR to switch the 9v.

The mains-powered relays are ( http://www.chauvetlighting.com/relay-pack ) a chauvet relay pack that have a 9-way input. Pin 5 is at least 9v, the other 8 are ground-to-enable connections. 

So, yeah, I could have a power-input to the relay controller board, and that would be easy. It's not quite so elegant though, and if I could get it to be bus-powered, I'd prefer it. Hence the questions :)


Cheers
   Simon

On Mon, 27 Aug 2012 17:08:49 -0700 (PDT)
Simon <g...@gornall.net> wrote:

> On Monday, August 27, 2012 4:37:05 PM UTC-7, Les Cargill wrote:
> > 
> > It might - it has a builtin charge pump. I would not trust one past 5V.
> 
> Hmm - not past 5v, huh...
> 
> Another option I thought of would be something like ( http://www.digikey.com/product-detail/en/VBT2-S5-S12-SMT/102-2078-1-ND/2295136 ) - although I'd have to put a 680R[*] resistor across the outputs to make sure there's always a 10% load on them even if no relays are switched on. Wasting power like that just seems .... wrong :) 
> 
> I'd assume that this power would be sufficient: The bench PSU reports I'm pulling 0.1A @ 5V into the '7660 circuit when I have all 8 relays switched on, and the above part says its max output is 0.167A. That ought to be enough headroom.
> 
> [*] working it out as requiring at least 17mA of current at 12v, so
> 
>                 12 = (17/1000) * R
>             => R = 705 = ~680
>             and Power dissipated = ~0.2W 
> 
> 
> > 
> > But can't you add another relay at the PC end
> > with enough energy? That would save you having a wallwart at the
> > far end.
> 
> Yeah, the above isn't really as clear as I ought to have been. When I said "controller", I meant the USB interface, not the relay. The setup looks something like:
> 
> [embedded] <--USB--> [relay controller board] <--25' cable --> [mains-powered relays]
> [  system   ]
> 
> Where the embedded system only has a couple of USB port outputs, and the relay controller board is (currently) powered from one of those outputs. It's an AVR coupled with an FT232 and some transistors to allow the AVR to switch the 9v.
> 
> The mains-powered relays are ( http://www.chauvetlighting.com/relay-pack ) a chauvet relay pack that have a 9-way input. Pin 5 is at least 9v, the other 8 are ground-to-enable connections. 
> 
> So, yeah, I could have a power-input to the relay controller board, and that would be easy. It's not quite so elegant though, and if I could get it to be bus-powered, I'd prefer it. Hence the questions :)
> 
> 
> Cheers
>    Simon

I've tried exactly that trick with a MAX232.  The charge pump is pretty
pitiful, even with the outputs unloaded I seem to recall only getting
about +8V out, and it dropped pretty sharply as you weighed it down.

Could just roll your own charge pump with diodes and caps, then just
have the micro generate a PWM signal into some beefy driver like an
LVC3G34, or just a transistor.  You'd want to use big caps to keep the
droop low, but it seems like you're not talking about all that much
current.

-- 
Rob Gaddi, Highland Technology -- www.highlandtechnology.com
Email address domain is currently out of order.  See above to fix.

> 
> Could just roll your own charge pump with diodes and caps, then just
> 
> have the micro generate a PWM signal into some beefy driver like an
> 
> LVC3G34, or just a transistor.  You'd want to use big caps to keep the
> 
> droop low, but it seems like you're not talking about all that much
> 
> current.

So, the AVR (an ATMEGA168A) can source up to 40mA per pin (with a total of 200mA per device) so I don't think I'd need the driver IC, but while googling for appropriate charge pump circuits, I ran across the LTC1263. This looks as though it can provide up to 90mA at 12v from a 5v supply, which ought to be plenty! I think I'll get me a couple of these to try breadboarding with :)

Simon

Simon wrote:
>>
>> Could just roll your own charge pump with diodes and caps, then
>> just
>>
>> have the micro generate a PWM signal into some beefy driver like
>> an
>>
>> LVC3G34, or just a transistor.  You'd want to use big caps to keep
>> the
>>
>> droop low, but it seems like you're not talking about all that
>> much
>>
>> current.
>
> So, the AVR (an ATMEGA168A) can source up to 40mA per pin (with a
> total of 200mA per device) so I don't think I'd need the driver IC,
> but while googling for appropriate charge pump circuits, I ran across
> the LTC1263. This looks as though it can provide up to 90mA at 12v
> from a 5v supply, which ought to be plenty! I think I'll get me a
> couple of these to try breadboarding with :)
>
> Simon
>

40 mA @ 5 V is still only 0.200 Volt-Amps - assuming the PIO isn't 3.3
or 1.x volt...

--
Les Cargill

On Monday, August 27, 2012 6:22:30 PM UTC-7, Les Cargill wrote:
> Simon wrote:
> 
> >>
> 
> >> Could just roll your own charge pump with diodes and caps, then
> 
> >> just
> 
> >>
> 
> >> have the micro generate a PWM signal into some beefy driver like
> 
> >> an
> 
> >>
> 
> >> LVC3G34, or just a transistor.  You'd want to use big caps to keep
> 
> >> the
> 
> >>
> 
> >> droop low, but it seems like you're not talking about all that
> 
> >> much
> 
> >>
> 
> >> current.
> 
> >
> 
> > So, the AVR (an ATMEGA168A) can source up to 40mA per pin (with a
> 
> > total of 200mA per device) so I don't think I'd need the driver IC,
> 
> > but while googling for appropriate charge pump circuits, I ran across
> 
> > the LTC1263. This looks as though it can provide up to 90mA at 12v
> 
> > from a 5v supply, which ought to be plenty! I think I'll get me a
> 
> > couple of these to try breadboarding with :)
> 
> >
> 
> > Simon
> 
> >
> 
> 
> 
> 40 mA @ 5 V is still only 0.200 Volt-Amps - assuming the PIO isn't 3.3
> 
> or 1.x volt...

I think I'm missing your point. The LVC3G34 has an output drive of 24mA, according to its data sheet. How does it help to put a 24mA driver chip in front of a 40mA pin ? 

I'm assuming here that the 2n2222 transistors are always going to be used to switch the 10v supply, so the driver chip would only really be useful if the AVR driving capacity was too low to usefully drive the transistors. 

Unless I'm missing something. Usually when I don't understand what someone's saying on here, I'm missing something [grin]

Simon.

On Mon, 27 Aug 2012 16:11:38 -0700, Simon wrote:

> I'm trying to hang an 8-way relay-driver off a USB port, with the
> circuit being bus-powered.
> 
> The relays in question need at least a 9v signal, and they can be at the
> end of a 25' wire (DB9<-~~->DP9). This part is fixed because it's an
> off-the shelf part.
> 
> The first attempt was to try using an ICL7660 (since I had them lying
> around) in voltage-doubling configuration. That worked fine until I
> actually wanted to use the relays [grin]. The 7660 could produce 9v with
> no load, but as soon as I connected just one of the relay pins, the
> voltage dropped to ~4v and the relay failed to switch.
> 
> So I was wondering if the MAX232 would do any better - I know it's not
> what it's supposed to be used for, but all I really want is +10v from
> the USB's +5v that will drive 8 relays... I'm supposing that the RS232
> drivers would probably be specced to drive the voltage down a long
> serial wire anyway, so maybe they'll do better ?
> 
> If it comes down to it, I can have a separate power-input to the
> controller (and basically make it not bus-powered) and then there's no
> problem, but if there's a good way of avoiding that I'd like to know
> about it :)

What's the resistance of the relay coils?

If you're willing to go surface mount anyway, you might want to consider 
using a boost regulator -- generating a few hundred mA at 9V from 5V 
should be easy.

-- 
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?

Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com

On Monday, August 27, 2012 7:10:34 PM UTC-7, Tim Wescott wrote:
> 
> What's the resistance of the relay coils?
> 
I don't have that to-hand, I'd have to measure them. Previously all I've done is have the non-bus-powered approach and a 9v wall-wart connected via a 2n2222 to the AVR. That worked fine :)

> 
> If you're willing to go surface mount anyway, you might want to consider 
> using a boost regulator -- generating a few hundred mA at 9V from 5V 
> should be easy.

Are there any that don't need so many support components ? It's been a while since I used one of those, but I seem to recall needing inductors and capacitors, sometimes resistors as well, according to the 'example circuit' they usually have in the datasheet... 

What was attractive about the LTC1263 was the power output (90mA) with only 4 caps (2x 0603, 2x0805). Of course, it's relatively expensive in small quantities (~$7) so if there's a better (read: can supply the power and has a small BOM requirement) option, I'm all ears :)

Cheers 
    Simon

Simon wrote:
> On Monday, August 27, 2012 6:22:30 PM UTC-7, Les Cargill wrote:
>> Simon wrote:
>>
>>>>
>>
>>>> Could just roll your own charge pump with diodes and caps,
>>>> then
>>
>>>> just
>>
>>>>
>>
>>>> have the micro generate a PWM signal into some beefy driver
>>>> like
>>
>>>> an
>>
>>>>
>>
>>>> LVC3G34, or just a transistor.  You'd want to use big caps to
>>>> keep
>>
>>>> the
>>
>>>>
>>
>>>> droop low, but it seems like you're not talking about all that
>>
>>>> much
>>
>>>>
>>
>>>> current.
>>
>>>
>>
>>> So, the AVR (an ATMEGA168A) can source up to 40mA per pin (with
>>> a
>>
>>> total of 200mA per device) so I don't think I'd need the driver
>>> IC,
>>
>>> but while googling for appropriate charge pump circuits, I ran
>>> across
>>
>>> the LTC1263. This looks as though it can provide up to 90mA at
>>> 12v
>>
>>> from a 5v supply, which ought to be plenty! I think I'll get me
>>> a
>>
>>> couple of these to try breadboarding with :)
>>
>>>
>>
>>> Simon
>>
>>>
>>
>>
>>
>> 40 mA @ 5 V is still only 0.200 Volt-Amps - assuming the PIO isn't
>> 3.3
>>
>> or 1.x volt...
>
> I think I'm missing your point. The LVC3G34 has an output drive of
> 24mA, according to its data sheet. How does it help to put a 24mA
> driver chip in front of a 40mA pin ?
>

I hadn't looked yet. I am not sure what the LVC3G34  would
do for you.


> I'm assuming here that the 2n2222 transistors are always going to be
> used to switch the 10v supply, so the driver chip would only really
> be useful if the AVR driving capacity was too low to usefully drive
> the transistors.
>

Sounds like a plan then. I was simply estimating the AVR driving 
capacity at a ( seemingly inadequate ) .2 V-A, and simply hanging
that on the FLASH Vpp driver chip would not have changed that capacity.

If you have a nice three-legger ( the 2n2222  ) driving the Vpp chip (
the LTC1263 ) then I expect you are golden.

Sadly, the elegance is lessened, but I'd take adequate current
over elegance.

> Unless I'm missing something. Usually when I don't understand what
> someone's saying on here, I'm missing something [grin]
>

If you got it figured out, don't pay any attention to me - all I
have is pieces of the thing on this end! :)

> Simon.
>

--
Les Cargill