How do I benefit from the +/- 40 V SoftSpan mode?

Hi CodyGray ,

the AD4858 has differential inputs, and the ranges relate to the differential inputs voltages (VINx+ − VINx−), not on the absolute ground related inputs voltages.

As an Example:

VIN+ = 20 V (against ground)

VIN- = -20 V (against ground)

VDIFF = VIN+ - VIN- = 20 V - (-20 V) = 40 V

And the other way round:

VIN+ = -20 V (against ground)

VIN- = 20 V (against ground)

VDIFF = VIN+ - VIN- = -20 V - 20 V = -40 V

And that is how you get a +/-40 V input range.

Just remember that you have do regard the input voltage as the difference between the +&- inputs pins, not as an absolute voltage of a single pin against ground.

regards

Hi, CodyGray .

CodyGray said:

First, I just want to double-check whether that is a correct understanding.

Your understanding is correct. To fully accommodate the Absolute Input Voltage Range, you need a supply of VCC with at least 3.2 V of headroom. Rehfi’s explanation is also accurate: if your Softspan is ±40 V, your signal amplitudes should be ±20 V, resulting in a total span of 40 V (from the difference of +20 V to -20 V).

To add on the previous explanations, let’s shift the perspective from focusing on the benefit of a ±40 V softspan to considering the advantage of using VCC = 24 V and VEE = 24 V. This approach gives you a much more flexible common-mode voltage (VCM). For example, with a softspan of ±10 V, if your zero crossing is at 0 V, you only need an amplitude of 5 V to achieve the full ±10 V range. This flexibility can simplify design and improve signal handling (see condition 1 from the image given below).

Going back to its benefit, as you can see at the second condition, if your VCM is at 15V, you can still achieve an input voltage range of up to 20V since your VCC is 24V. The 20V resulted to VCC-3.2 (24 - 3.2 = 20.8V).

Another standout feature of the AD4858 is Seamless High Dynamic Range (SHDR). When enabled, the converter automatically adjusts gain based on each sample’s differential voltage. High-magnitude signals use the standard gain, while lower-magnitude signals receive increased gain, reducing input-referred noise and improving dynamic range.

Though, if you are limited with your supply, just like you mentioned by comparing it to ADS9813, then you can try looking into AD7606C-18.  

The AD7606C-18 operates from a single 5 V supply and accommodates the following input ranges through its Programmable Gain Amplifier (PGA) and when sampling at throughput rates of 1 MSPS for all channels:

• Bipolar single-ended: ±12.5 V, ±10 V, ±6.25 V, ±5 V, and ±2.5 V

• Unipolar single-ended: 0 V to 12.5 V, 0 V to 10 V, and 0 V to 5 V

• Bipolar differential: ±20 V, ±12.5 V, ±10 V, and ±5 V

Regards,
Jo

Thanks NathanT and rehfi for your answers!

But I'm still struggling to understand why the SoftSpan range is called ±40 V when it can only accommodate input signals that range from -20 to +20 V. To me, that would be a range of ±20 V.

 NathanT writes:

if your Softspan is ±40 V, your signal amplitudes should be ±20 V, resulting in a total span of 40 V (from the difference of +20 V to -20 V).

I would call that a SoftSpan of 40 V, where the total range (difference) is 40 V (from -20 V to +20 V). In fact, there is a SoftSpan setting for 40 V, but then there is also a SoftSpan setting for ±40 V, so I would have expected those to be different things. With a SoftSpan setting of ±40 V, why can't I have an input signal that is ±40 V? This is very unintuitive to me!

But this seems to be exactly what you are saying:

with a softspan of ±10 V, if your zero crossing is at 0 V, you only need an amplitude of 5 V to achieve the full ±10 V range.

A bipolar signal that ranges from -5 V to +5 V (i.e., ±5 V) has a range of 10 V, but not a range of ±10 V. You show exactly this in your drawing. It does not matter where the common-mode voltage is. In other words, it doesn't matter whether the bipolar signal is centered around 0 V or some other arbitrary level. The total amplitude is 10 V peak-to-peak (Vpp), but it's referred to as a ±5 V. Right? What am I missing here?

My common-mode voltage will be 0 V, so being able to accommodate shifts in the input signal is not advantageous to me—although I certainly understand the general benefits of that and how it might be useful in other applications.

I also am very interested in that SHDR feature. That and the ± 40 V range are the two reasons why I was strongly considering the AD458x series over similar parts, whether from TI (as I had mentioned) or from ADI (like the AD7606C-18 that you also mention).

Speaking of the AD7606C-18, its datasheet seems to be consistent with my interpretation. For example, see Table 11, the bipolar input voltage ranges. For a differential, bipolar signal that is ±20 V, it says that the PFS is +20 V and the NFS is -20 V, with a mid-scale of 0 V. That is 40 V peak-to-peak, but ±20 V. Why does this same nomenclature not apply to the AD485x?

Best,
Cody

Hi CodyGray ,

CodyGray said:

Speaking of the AD7606C-18, its datasheet seems to be consistent with my interpretation. For example, see Table 11, the bipolar input voltage ranges. For a differential, bipolar signal that is ±20 V, it says that the PFS is +20 V and the NFS is -20 V, with a mid-scale of 0 V. That is 40 V peak-to-peak, but ±20 V. Why does this same nomenclature not apply to the AD485x?

When you look closely, the nomenclature for the AD4858 is exactly the same, compare the AD4858's "Table 12. SoftSpan Range Properties" with the AD7606C's "Table 11. Bipolar Input Voltage Ranges".

CodyGray said:

A bipolar signal that ranges from -5 V to +5 V (i.e., ±5 V) has a range of 10 V, but not a range of ±10 V.

This is true for a bipolar single-ended signal, where a single input is measured against GND. The AD4858 is a differential input ADC, where two inputs are measured against each other. One can be GND, but can also any other value, so you must take this into account when describing (and naming) ranges. And then we are back to the example from my last post.

When you need a ±40 V single-ended bipolar input range, then the AD4858 won't work, and I think there are not many ADCs that support such ranges nativley. But can work around it, i.e. by increasing the input range with external resistors. There is a nice appnote from TI (sbaa244), which is very easy to implement with the AD7606C.

At my lab this whole single-ended vs. differential vs. absolute input voltage thing evolved into some kind of internal joke, it is easy to misunderstand, mostly not explained very well, and you have to get your head around it before it gets better.

Regards