ASMOP: relays vs. CMOS switching

[Leandro]

Inspired by GEOFEX's ASMOP design, I'm planning on building something very similar.  Doing some initial parts research with local electronic components stores, I was somewhat disheartened by the price of DPDT relays (my design uses 9 DPDT's).

I made the following analysis, based on a few assumptions about the parts needed:

Using DPDT small signal relays
- 9 x relays
- 9 x 1N4148's
- 9 x 2N3904's
- 9 x 0.1 uF non-polarized capacitors
- 9 x 1 kohm resistors

Using CMOS switching (CD4053)
- 6 x CD4053's
- 18 x 2.2 uF electrolytic capacitors
- 63 x 1 Mohm resistors

A few calculations indicate that the second alternative will come in at under half the cost.

What do you guys think would be the drawbacks (apart from the increased parts count) of using CMOS analog switches instead of small signal DPDT relays?

[12afael]

you need to bias the signal to pass it though cmos more caps more resistors and cmos have a internal recistance of about 80 to 150 ohm .
you need a good power supply of +-15 v for a low resitance.

I try with 4016 and 4066 and I don´t like them

[R.G.]

I recommend the NEC EA-2 Series of relays. They're about $2.50 each from Mouser in the USA. Don't know about other places. CD4053's are always going to be less expensive than relays.

[Leandro]

I recommend the NEC EA-2 Series of relays. They're about $2.50 each from Mouser in the USA. Don't know about other places. CD4053's are always going to be less expensive than relays.

Is CMOS switching not recommendable, then?

I guess I'll go with the relays.  Should I get the 12 VDC or 5 VDC types?  Finally, non-latching, single coil latching or double coil latching?

Thanks, guys!

[R.G.]

Is CMOS switching not recommendable, then?

No. Relay switching has one set of virtues; cheap price is not one of them. CMOS switching has other virtues. They are just different, and neither is perfect.

Relays: High off resistance, low on resistance, good isolation from switching control signal, high power usage or complicated circuits for latching, moderately expensive

CMOS: High off resistance, moderate on resistance, fair isolation from control signal, low power usage, simple control circuits, low cost.

I like CMOS switching, but you have to know and work with its set of needs. CD405x chips have to have the DC bias at the input/output pins be half the power supply voltage on the chip to minimize switch pop from the control signal.

Relays have to have the control signal (coil voltage) slowed down to avoid coupling switching pop into the audio path.

[Leandro]

Thanks, R.G.!

Excuse my insistance, but I'm still green on this stuff...  Should I use 12 V or 5 V relays?  Logic seems to indicate that 5 V relays are right on for this, because that's pretty much the standard output on-voltage of logic devices (I assume this for PICs and AVRs).  However, most schematics and articles I've seen use 12 V relays which draw their "on" voltage from a 9 V supply (when a few datsheets I've looked at guarantee that 12 V relays will switch at 8.4 V, which is probably too close for comfort).

Finally, the switching type: non-latching relays are bound to drain a lot of current when left in the "on" position for long sets or practice time.  I won't be using batteries, so it's not really the power consumption that bothers me, but the feeling that the lifespan of the relay will be dramatically shortened in this case (I read a datasheet that said so).  However, non-latching relays are the easiest to understand and use.

Can you recommend something on these two issues from your experience?

[12afael]

I suspect single coil latching should work.

Should I use 12 V or 5 V relays? Logic seems to indicate that 5 V relays are right on for this, because that's pretty much the standard output on-voltage of logic devices (I assume this for PICs and AVRs).

the ic don´t give you enough amperage to the relay so you need a trancistor.

CD405x chips have to have the DC bias at the input/output pins be half the power supply voltage on the chip to minimize switch pop from the control signal.

why you need to bias the output? I never understand that.
if you decoupled it with caps don´t need bias the output. and if you will in to other cmos it is already biased or not???

[Brett Sinclair]

Hello everybody!
First post...  

What about these Maxim IC CMOS switches?

http://www.maxim-ic.com/quick_view2.cfm?qv_pk=2054
http://www.maxim-ic.com/quick_view2.cfm?qv_pk=2055

I have a few of them lying around but need to experiment a bit. I'm not a techy, but what do you guys think about the specs?

cheers!

[R.G.]

why you need to bias the output? I never understand that.
if you decoupled it with caps don´t need bias the output. and if you will in to other cmos it is already biased or not???

The chip works fine without the input/output pins being biased - but it introduces switch pop.

The CMOS switches consist of a pchannel and an nchannel device. When these are balanced in the middle of the power supply by biasing, the pop from the p side is equal and opposite to the n side, and almost no pop is induced.

[Leandro]

Should I use 12 V or 5 V relays? Logic seems to indicate that 5 V relays are right on for this, because that's pretty much the standard output on-voltage of logic devices (I assume this for PICs and AVRs).

the ic don´t give you enough amperage to the relay so you need a trancistor.

Yeah, you're right, I can't drive the relays with an IC output pin...  :oops:

However, since the whole project is IC's, relays and LEDs (i.e., no signal processing), I could use 5 V relays and keep the design simple with a single supply of 5 VDC, right?  Or will this not be enough for the relay driver circuit?

[Leandro]

What about these Maxim IC CMOS switches?

http://www.maxim-ic.com/quick_view2.cfm?qv_pk=2054
http://www.maxim-ic.com/quick_view2.cfm?qv_pk=2055

Hey, those are impressive!  Guaranteed maximum on resistance of 2.5 or 5 ohms, versus a maximum of 1k-1.3k in the case of CD4053's!

[R.G.]

Guaranteed maximum on resistance of 2.5 or 5 ohms, versus a maximum of 1k-1.3k in the case of CD4053's!

Hmmm... let's see... switch drives an amp input of about 1M. So the voltage lost to the switch resistance is 2.5 ohms/1M+2.5 ohms in one case and 1.3K/1.3K +1M in the other case. For a 1V signal, that's a 2.5uV loss versus a 1.3mV loss. In the Maxim case, you have 0.9999985 volts of signal left, and in the 4053 case, you have 0.9987 volts of signal left.

I wouldn't pay extra for that difference in performance if that's all I get.

For the Maxim chips, the important thing is over on the right hand side of the datasheet - charge injection. All CMOS switches inject some charge. That's the basis for my comments on why bias. Charge injection is minimized for CD405x family when the in/out pins are held at the middle of the power supply. How low do the 405x go? I don't know - it isn't specified.

However, for the maxim switches, you get 300pC of charge injected when the thing switches. From the fundamental equation that Q=C*V, we see that the voltage change from 300pC is 300pC/Capacitance. In this case, it's the capacitance of everything following the output switch. For a 10 foot cord with 30pF per foot, you get 300pF of capacitance, and V = 300pC/300pF = 1V.

Ack! That's horrible! A 1V switch pop? Maybe.

The thing is, the switch can't inject that all at once, and the input resistance of the following device eats some of the charge as the cable capacitance is charging up. And the 300pC value is a typical value - which is by the way a worthless specification.

There is another Maxim family that offers 3pC charge injection, a factor of 100 lower. The 4617/18/19 family is intended for voltage mode switching. The 466x family is intended for current-mode switching, which explains the difference.

All are pin compatible with the CD4053, so maybe the thing to do is to put in a socket and substitute away.

But getting low pop is not always straightforward.

[Leandro]

Guaranteed maximum on resistance of 2.5 or 5 ohms, versus a maximum of 1k-1.3k in the case of CD4053's!

[Long story short: they're not really that impressive   ]

R.G. just gave me flashbacks of my electrical theory courses at my university.  Wish I had retained more of that...   :oops:

Anyway, thanks for clearing that up.  I'd decided to use relays, anyway.  They may be more expensive, but it's not like I'm going to have to file for bankruptcy because of them, and it's bound to be a darn fine learning experience.  That being said, I think I'm ready to take the next step and finalize the schematics, but prior to that I'd really appreciate it if someone could confirm my following assumption:

Yeah, you're right, I can't drive the relays with an IC output pin...  :oops:

However, since the whole project is IC's, relays and LEDs (i.e., no signal processing), I could use 5 V relays and keep the design simple with a single supply of 5 VDC, right? Or will this not be enough for the relay driver circuit?

[R.G.]

5V will work fine for driving the relays (the ones with the 5V coils), including working with driver transistors. Be aware that you ought to add up the number of relays that will be on at any one time and make sure your power supply can provide that much at the same time as powering your logic.

[Leandro]

5V will work fine for driving the relays (the ones with the 5V coils), including working with driver transistors. Be aware that you ought to add up the number of relays that will be on at any one time and make sure your power supply can provide that much at the same time as powering your logic.

Yeah, that's another thing I'm worried about...

The relays I'm going to get have a maximum switching current of 2 A!  There's no steady-state on-current data in the datasheet, but since they're non-latching, I guess it won't be trivial.  Since at most 8 relays will be simultaneously switched on, in true rigor I should have a supply rated for 16 A spikes (excluding IC's, LED's, etc.), right?!

I'll have to do some research on that.  One of our Danish co-forumers has a link from his own web site to the Delta Audio site, which hosts a pretty elaborate +/- 15 VDC 3 A power supply schematic.  Bipolar voltage notwithstanding, it looks like I need that supply on steroids...

[niftydog]

The relays I'm going to get have a maximum switching current of 2 A!

Uh, I think that's the rating of the switch contacts. What you want to know about is how much current the coil consumes when energised.

I don't think you'll be needing an 80W power supply!!!!

[Leandro]

The relays I'm going to get have a maximum switching current of 2 A!

Uh, I think that's the rating of the switch contacts. What you want to know about is how much current the coil consumes when energised.

I don't think you'll be needing an 80W power supply!!!!

Here's a link to the relay's datasheet.  The switch contact rating is on there, too, but I quoted a different parameter: the "switching current".  I realize nonetheless that I must be making some mistake, because an 80W power supply is impractical and not often heard of in this context.

So I guess 2 A * 8 relays is a wrong estimate...   :roll:

This forum is probably the one place where I actually like to say stupid things and make a fool out of myself -- people always seem eager to discuss and get each other pointed in the right direction.