Multiplexer - Deriving Cds & Switch OFF resistance value from OFFisolation Vs frequency curve

Hi Jebas,

Sorry for the delay in replying.  We are looking at your request now and will post comments asap. 

Regards,

Eric.

Hi Jebas,

For Cds we usually estimate using the following relationship, Cds = (Cs(off) + Cd(off)) - (Cd(on) or Cs(on)).

On the Roff value.  Can you provide more information on what your issue is in your ADC application?  What frequency are you working at?  Are you trying to estimate it at DC?   

Regards,

Eric.

Hi Eric,

Thanks for the response.

On Calculation of Cds:

From datasheet, Cs(off) = 8pF ; Cd(off) = 40pF ; Cd(on) = 50pF

So, According to your estimation;

                                        Cds = 8 pF + 40 pF - 50 pF = 2 pF

I have following questions on this;

1. Is this estimated Cds value, gives the addition of all 16 switches Cds present inside the IC. i.e When All inputs are connected to a single source and EN=Low, all switches are OFF and so that capacitance Cds comes into parallel.

2. If above said statement is correct then, Cds across a single switch will be around 125fF (2pF/16). It matches with my initial calculation. Please do verify this.

On Roff Value;

I am working on 16bit resolution differential DAS. The signal frequency is almost DC, 200Hz. Since, there are multiple channels (24 channels in my application), OFFisolation become the major parameter for me. After going through the datasheet, at Lower frequencies the OFFisolation curve is flat and the value equals to the ratio of Roff and RL. So, I calculated the value of Roff as in my previous post. I had a doubt in this value because, While choosing a MUX for my application I came across so many other vendors parts also. But, I found good OFFisolation in your parts only.

So, I want to know is there any special technology or architecture which makes "Analog Devices" unique.

I came across LC2MOS technology of your Company. Is it helping to produce such a good devices. If so, please direct me towards a good document related to this technology.

Thanks,

Jebas.

Hi Jebas,

Your estimation of 2pF is correct. When the on capacitance measurement is made the channel to be measured is the only channel in the on state, all other channels are off (high impedance). The measurement signal used to calculate the capacitance value is a 1Vpp 1MHz sine wave.

The capacitance values in the data sheet relate to the capacitance to ground. Using the equation above will give you the channel capacitance to ground not across the channel. I believe your method in the first post will give you the best estimation of the channel capacitance.

Regards,

Joe

Hi Joe & Eric,

Thanks for your responses.

As I said before my DAS system has;

24 channels [ it is realized using three ADG407 ]

16 bit resolution

input signal frequency of 200Hz

Differential operation

My calculation shows that ADG407 supports 16 bit resolution.

OFF isolation:

Roff_single switch = 1.33 Gohm [ For confirmation I have send queries to Analog devices]

24 channels - Attenuation @200Hz = -110.21dB (single ended) ; with RL =200ohm

* 24 channels - I used three 8:1MUX (ADG407) in parallel to realize 24:1 MUX.

Cross Talk :

Assumption made is, AAF output impedance is 50 ohm (Low Impedance)

24 Channels - Cross Talk @200Hz = -117.78dB (single ended);

Charge Injection:

In ideal conditions, Differentially Charge injection is zero.

Break Before Make time :

This is the mandatory time duration between switching of channel.

Since, the output of MUX has only capacitance of [ Cd(off) - Cin(IA) ], any dc current during this period can develop voltage across this capacitance.

The only DC current during this period is Id(off),

Drain leakage current, which is equal to 20nA(max).

From datasheet, Break Before Make time is 10ns typical.

==> Voltage developed = I * T /C = 20nA * 10ns / 50pF = 4uV

Please confirm that the selected part ADG407 supports 16 bit resolution.

Thanks,

Jebas.

Hi Jebas,

I think in general the ADG407 is a reasonable starting point but in a system like this I feel achieving 16-bit accuracy is linked to a lot of factors and not just the switch.  You should talk to DAS support as well.  I notice that the inputs to the ADC channels are high impedance so parameters like switch charge injection may still need to be considered as it would affect your settling time.  Note, the newer ADG1207 has a lower and flatter (w.r.t. voltage range) charge injection response than the ADG407.  Therefore it maybe worth considering the ADG1207 as well.  My feeling is switch leakage, charge injection and switching time will all be factors affecting how you reach a 16-bit accuracy.  So in summary, I think using these switches is a good starting point for your application.

Regards,

Eric.

Hi Eric,

Thanks for your response. The discussion with You and Joe is much useful.

I have another one doubt to be cleared.

The first figure taken from ADG407 data-sheet.

Whereas, the second figure taken from ADG1207 data-sheet.

While comparing these two images, the one coressponds to ADG407 looks more ideal and gives a impression that it might not be from practical measurement.

Can I go with this data given in the datasheet of ADG407.

Also, as you said in the previous post, we are little worried about charge injection with ADG407 even our architecture is differential. Thanks for suggesting ADG1207.

But, ADG407 overlooks ADG1207 interms of OFFisolation, Crosstalk, Rdson etc.,. So, If I am using ADG407 with bottom plate  sampling inorder to discard the effect of charge injection, Whan can be your command on this.

Thanks,

Jebas.

Hi Jebas,

The ADG407 is an older part and it looks like the off isolation plot was made with some measurements that were then extrapolated for the plot. The ADG1207 off isolation was measured on a network analyzer and so full measurement data is available and the low frequency values in the plot show the equipments noise floor as opposed to the performance of the part.

At the frequencies you are operating I would think that there would not be much difference in real isolation between the ADG407 and the ADG1207. The only advantage for the ADG407 is that it is a physically larger part and so there is more separation between channels but at 200Hz this should not create a significant difference between the ADG1207 and the ADG407.

If you are ok to accept the higher charge injection of the ADG407 and the timing specifications are acceptable, this part should be suitable for the applications you have suggested.

Regards,

Joe