Applying control voltages to 2600-RBT-200 and 2650-RBT-3K (CVU-3K-KIT) with an external bias-T

This is a continuation of the thread in https://my.tek.com/en/tektalk/components/ea75857c-7640-ef11-b4ac-6045bd03a1eb.

The RBT-200 / RBT-3K require a +- 10V bias voltage applied to the AC input to close the AC switches, see CVU-3K-KIT user manual.
Normally, this control voltage is provided by the DC bias of the CVU itself. However, this has a major drawback: When a typical CVU performs an open correction, the DC bias is turned off and the open correction is performed with the AC switch open. Therefore, one needs to do a "manual open-correction", measuring the impedance for different frequencies and subtracting it offline (which is how Keithley ACS does it, see discussion in previous thread).
This can be quite cumbersome, and if the CVU can only source a voltage on the CVUHI side (and not CVULO), only one bias tee switch can be closed.

In this thread I want to ask if it is possible to use a second, external bias tee to provide the +- 10V to the RBT bias tees.
This would allow for controlling the bias tees independent of the state of the CVU and even to perform a "normal" open correction. Footnote 4 of the CVU-3K-KIT user manual states that for the "AC guard" mode, the +12V control voltage "Must be supplied by an external power source". Is there any more information on how one would use an external power source?

At low frequencies (100 kHz) it seems quite challenging to design a bias tee with a high enough isolation between AC and DC, especially if the DC resistance needs to be low enough such that the control signal of +-10V can be maintained within 250 mV.

Any information or recommendations would be greatly appreciated!


 

Hi,
Are you working with only two terminal devices?  In your original post it seems that might be the case.
Do you need to use more than one test frequency?

As you have discovered, when using bias tee, the mode needs to be managed during collection of the compensation values.  For nF and smaller, open compensation is most important.
You can accomplish that with just subtracting the open ckt measured value from your DUT measurement (at same test frequency).

The KXCI commands for collection and application of the compensation to measurements is implementing the same functions that Clarius gives for collection and use of compensation.  However, these are not expecting the bias tees.

Most of our 3KV HVCV installations are for three terminal devices (Ciss, Coss, Crss).  For this, we need different compensation values for each of the three measurements (due to cabling differences).  This influenced the ACS software implementation approach.

I have concerns about stacking up bias tees to address the issue.

 

Hi Andrea, thanks for your reply!

Yes, we are measuring 2-terminal devices (particle detectors based on SiC p-n junctions with depletion voltages > 1kV), with capacitances ranging from 0.1 pF to 100 pF. Normally, we are just using a single test frequency around 100 kHz (and only the open correction). For the foreseeable future we are not going to perform 3-terminal measurements.
I understand that our application is a non-standard, but power electronics equipment seems to be the only niche offering measurement solutions up to 3kV. We obtained a CVU-3K-KIT for our C-V measurements because the cost of the triaxial 3kV connectors places it at a similar price point as building the bias tee ourselves.

The open correction procedure done by ACS (or as we implemented it in our own software, subtracting the impedance measured with lifted needles) is working so far with a K4200A-SCS. First I was under the impression that the connection compensation data was stored in the CVU internally, but if this is not the case, I understand now the implementation in Clarius / ACS.

At our institute we are sharing a K4200A between labs, so it is not always available. Being able to use a "regular" LCR meter (for example an E4980A) would be preferable. More specifically, I want to ask if it possible to use the CVU-3K-KIT bias tees with an LCR meter that has no internal bias tee (or a bias tee just for CVUHI, with CVULO fixed at 0V)?
I share you concerns with stacking bias tees, but I do not see any other way to obtain the +-10V control voltage if the LCR meter can not provide it on CVULO.

If this is not possible to implement, we will keep using the K4200A CVU for now.