Is isolation my solution? Advice requested

Hi MTJones,

The first thing I would check is the absolute voltage levels at the ADG725 switch inputs. The voltage drop across the shunts may be small since the resistors are of high value, but the actual voltage must be within the VDD supply range of the ADG725. Since you're dealing with 60 V DAC outputs, and the shunts are measuring high-side, this must be considered.

Stephen_Nugent do you have any other suggestions?

DaveC

Thanks for the reply Dave.

The absolute voltage levels at the multiplexer switch inputs are sub 100 mV in the testing that I have been doing. 

I am not measuring (high side) the shunt between the high voltage DAC output and the resistor network.  I am always measuring (low side) across the shunt between the grounded DAC output and the resistor network. 

I am not sure if this will provide any additional insight into the problem, but I guess more information can't hurt.

When I set a DAC output to ground it is not true ground.  I typically measure a 1-2 mV difference between the ground pin of the USB port on the breadboard and the ground pin of the DAC.  The difference between the ground pin of the DAC and a DAC output set to ground is typically around 30 mV.  I measure 30 mV at the DAC output set to ground independent of whether anything is connected to the output or not.  When I set a DAC output to float I will typically measure 170-180 mV at the output when nothing is connected to the output. 

When I first wired the circuit I kept analog and digital grounds separate.  Since then I have tied them together since the wiring is a bit easier that way and, as far as i can tell, it does not seem to affect anything.

As part of the 60 V boost converter, I have two 68 uF capacitors and a 330 uH inductor.  Perhaps I should be looking at isolating the boost converter from the rest of the circuit instead of the DACs and multiplexer?

Hi,

You have to be careful that the part does not exceed the absolute max ratings at any time, even during power up. If there is a voltage present on the analog input that exceeds VSS - 0.3V or VDD + 0.3V, damage can be caused to the device. This means that is it very important that VDD is present before a signal is applied to the analog inputs.

Also you need to be aware of the continuous and instantaneous current that is flowing through the switch. The ADG725 can handle a continuous current of up to 30mA. Also it can handle a peak current of up to 60mA for a very short period of time. You need to be careful it is not exceeding this current limit when you are switching between channels.

Exceeding either of these, or any other of the absolute max ratings, could cause device failure issues.

Regards,

Stephen

Looking at the 60V domain is the first thing I would do. I understand that you're taking a measurement (closing switches) when the corresponding DAC output is at ground. But you also need to consider that the switches may be connected (but open) to higher voltage DAC outputs when a measurement is not being made.

DaveC

Stephen Nugent wrote:

You have to be careful that the part does not exceed the absolute max ratings at any time, even during power up. If there is a voltage present on the analog input that exceeds VSS - 0.3V or VDD + 0.3V, damage can be caused to the device. This means that is it very important that VDD is present before a signal is applied to the analog inputs.

 

Only the digital section of the AD5504 is powered up initially so there is no input on the multiplexer until the microprocessor signals the DACs.  This will happen within the first second of power up, but I am not sure on the exact timing off the top of my head.  Still, I can't imagine that the DAC output states get set before VDD is present at the multiplexer. 

Stephen Nugent wrote:

Also you need to be aware of the continuous and instantaneous current that is flowing through the switch. The ADG725 can handle a continuous current of up to 30mA. Also it can handle a peak current of up to 60mA for a very short period of time. You need to be careful it is not exceeding this current limit when you are switching between channels.

I am usually measuring continuous currents from 0-50 uA with the maximum around 100 uA.  The maximum current that the circuit is designed to measure is 100 uA.  I am not sure about instantaneous, though I do sometimes see a spike when a switch is made.  

I have been assuming that this current spike is actually not a real spike.  The multiplexer is break-before-make so when switching between channels the input to the instrumentation amplifier is open.  My thought has been that these apparent current spikes occur when the microprocessor reads from the amplifier during a transition. 

DaveC wrote:

I understand that you're taking a measurement (closing switches) when the corresponding DAC output is at ground. But you also need to consider that the switches may be connected (but open) to higher voltage DAC outputs when a measurement is not being made.

Ok, I see your point now.  If this is the cause of failure for the multiplexer, then I need a new multiplexer.  Or I will need to dedicate some the the DAC outputs to be set to ground and have no direct connection between the multiplexer inputs and the high voltage DAC outputs.

DaveC wrote:

Looking at the 60V domain is the first thing I would do.

This is where I am beyond my depth.  Any advice for what I need to consider here or resources you can point me to?  The boost converter is a Unitrode UC2577-ADJ.

The absolute maximum supply voltage on the DAC is only 64V. It is quite possible for the DC-DC converter to exceed that and damage the DAC. This is especially true at low load currents where regulation might be poor. You need to monitor this line to make sure the supply is safe. Since your margins are fairly small you will need at least somewhat decent test equipment to take the measurements. However you could also just lower the supply to 50V and run your tests to see if the DAC survives to see if this is part of your problem.

Regarding the analog mux adding ESD type protection to the inputs might be a good option. Since you mentioned that the actual voltages during operration are only in the 100 mV range you could also simply place forward biased silicon diodes in parallel with the inputs. They will shunt almost no current at 100 mV but guarantee that the inputs will never see overvoltage. They won't help with voltages below ground though, so make sure that is not an issue...

Klaus

You could segregate the low-voltage circuit onto a separate ground domain as shown by the red box. Then you need isolators to communicate across the isolation barrier.

The digital communication between the DAC and uC is the easy part, as ADI has many devices tailored to various peripheral interfaces such as I2C or SPI. The tough part is the MUX, since you have dozens of analog signals. This will take a lot of area, and analog isolators may not have the accuracy and low noise characteristics that you require.

You could isolate the MUX and InAmp in the DAC domain, but that won't help if it's the DAC that's causing the damage.

I think you should figure out the root cause before trying to apply isolation to fix the issue.

DaveC

Thanks all for the reply's,

KRZ wrote:

The absolute maximum supply voltage on the DAC is only 64V. It is quite possible for the DC-DC converter to exceed that and damage the DAC. This is especially true at low load currents where regulation might be poor.

This is something I have been worried about from the beginning.  I often check the output of the boost converter with a multimeter.  After connecting power, it usually takes a moment for the output voltage to reach 61 V (I have it slightly over 60 to ensure I can get 60 out of the DAC), but after that the output is stable every time I check it.  Even when the DAC or mux fail, the output of the converter is always 61 V.  Now, since I am only checking it periodically with a multimeter, it could be spiking without me knowing it, but again I have never seen it above 61 V.  I will definitely continue to keep an eye on this.

KRZ wrote:

However you could also just lower the supply to 50V and run your tests to see if the DAC survives to see if this is part of your problem.

good idea.  The DAC seems to last longer than the mux, but operating at lower than the maximum makes since.

KRZ wrote:

Regarding the analog mux adding ESD type protection to the inputs might be a good option.

I don't know anything about this, so I will need to read up on ESD protection.  Do you have any suggestions for where to start?  Does a particular class or category of component come to mind?

KRZ wrote:

Since you mentioned that the actual voltages during operration are only in the 100 mV range you could also simply place forward biased silicon diodes in parallel with the inputs. They will shunt almost no current at 100 mV but guarantee that the inputs will never see overvoltage. They won't help with voltages below ground though, so make sure that is not an issue...

If I have a forward biased diode in parallel with the a high side shunt, wouldn't I short the high voltage DAC output to ground? 

As Dave pointed out, the mux inputs on either side of the high side shunts are seeing high voltages even though those inputs are not active.  If I change to dedicated grounds then I can eliminate this problem.  In this case the didoes will protect the mux inputs without causing a problem on the high side.  This is a good idea, thank you.