Analog Audio switch with 4053

[davidsrsb]

Still asking for the exact part number of the 4053s - some types are not +/- 5V eg 74LV4053

You should me able to put a multimeter from the inputs to the switch output while powered and measure the on resistance, which should be around 2k2 ie 2k + 2x100R

[kmt]

@Havoc:
A huge thanks for the explanation! Makes some things much
clearer now.

@davidsrsb:
sorry, I haven't been able yet to look at the circuit (I work
for a german broadcast station and had a lot unexpected work this week), but I can assure You that I didn't use
a 74..4053. Either HEF4053 or CD4053 or HCF4053. As I remember, I downloaded the datasheet from the manufacturer and everything seemed to be ok. I'll take my multimeter and have a look at the on resistance. I hope I'll manage that on wednesday
evening (if the NDR doesn't give me overtime :-)

Greetz
kmt

[kmt]

I had a short glimpse at the circuit:

The switches are MC14053 by Motorola, the OpAmps
are NE5532AP by Texas instruments. Th OPs are
connected to +/-15V, the switches are connected to
5V/0V/-5V.

@Havoc:
no, the fault occurs with the normal signal switched
"on".

Greetz
kmt

[VivaVee]

Something else to think about. The 4000 series analog switches have very poor off isolation - only about 50dB. This is improved if the load resistors are very low but then the distortion rises rapidly as the load falls.

If you have the opportunity to redesign the circuit (not always possible) and you want solid state switches then have a look at the Analog Devices SSM2402 and SSM2404. These have much greater off isolation (due to a series-shunt-series switch configuration) and much lower charge injection (you will likely notice an audible click each time the 4053 switch changes state)

Definitely install a resistor on the opamp input ( >10k to minimise loading/distortion). If you want to increase this resistor further then choose a pin-for-pin compatible opamp with much lower bias current. The NE5534 is still useful but its Ib specifcation is in the dark ages - which is when it was born!

These switches are also very susceptible to failure from voltage spikes on the power supply. And therefore, likely to exhibit strange behavious if 'merely' damaged by these spikes rather than actually destroyed.

I have spent just a little time removing all the 4000 series multiplexers from a large tester to imrpove performance and relaibility. I would NOT recommend thing for anythign but the most cost sensitive project.

[kmt]

Hi!

Perhaps I should really change the 4053. The problem
is, that I would have to change these in an existing circuit,
so I would have to search an ic with exactly the same pin assignment and voltage range. Does someone know
such an ic?

The off isolation isn't a problem in my circuit because
of the circuit design itself. Connecting the switches in
the way I did improves the off isolation for the "not used" input
dramatically.

The voltage spike problem could be an issue. I'll have to check
in my free time.

@VivaVee: do you mean a resistor in the signal path from
the switch to the OP, or a resistor from the OPs input to GND?

Greetz from Hamburg
kmt

[jcx]

http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1352

[kmt]

@jcx: aah, thanx! I'll have a look where to buy these...

Greetz
kmt

[VivaVee]

@VivaVee: do you mean a resistor in the signal path from
the switch to the OP, or a resistor from the OPs input to GND?

I mean from the OP input to ground. As suggested above, check out Maxim. They have improved versions of the 4051/2/3 that are plug-in replacements. They actually specify the charge injection which suggests that that characteristic is actually an improvement over the standard devices. The on-resistance is certainly much lower and better matched between channels.

Re: Maxim switches - You may be able to get samples from the Maxim website.

[davidsrsb]

MC14053 are ok at +/-5V. These are very old devices with limited ESD protection and some tendancy to latch up if the inputs are driven beyond the supply rails - I would have a good look at the two inputs.

Another thing to check is that the control input is <0.5 V low and >4.5V high, TTL levels are not enough for this device.

I consider your circuit to be properly designed to work even with the limited OFF isolation of a 4053 switch because of the series/shunt arrangement. The NE5532 should be OK in this appication, it's nothing special but good enough.

[uglydavegray]

i cant see the need to change the resistors as your c/d looks just fine to me