Products > Test Equipment

New MicSig/EEVblog DP10007 HV Differential Probe

<< < (32/38) > >>

Centmo:
I just bought two of these probes and am following your posts intently  :D

ExaLab:
Thanks for your kind feedback!

Getting straight to the point I can confirm that for frequencies lower than approximately 50MHz the anomalous degradation of the cmrr of this probe is mainly linked to the stray capacitances.
Let's analyze the CMRR behaviour in the case of an hypothetical imbalance between the two equivalent stray capacitances at the op-amp inputs (capacitances referred to the reference plane).
It can easily be demonstrated that for an ideal op-amp configured as in fig. 2, for C values of the order of a few pF and frequencies lower than 100MHz, the CMRR is very close to:

                                        CMRR = 20 Log (ω Rp ΔC)         (3)

Where:    ω = 2 π f
      Rp is the parallel of the resistors that relate to the node
      ΔC is the capacitance misalignment between the two input nodes

Note that in our device (DP10007), Rp is equal to:

               83 ohm   (x10   range)
               333 ohm   (x100 range)

From the above relationship (3), it can be noted that:

      1) In an ideal op-amp the CMRR degrades with a slope of 20db/decade
      2) In our probe, for a given ΔC, the CMRR of the x100 range is 12dB worste than the x10 range


Some of you may have noticed that at 10MHz the CMRR difference between the two ranges is greater than 12dB (e.g. in my specific case it is: 39dB – 22dB = 17dB ).
The reason for this is very simple: the x100 range has a cap imbalance higher than that the x10 range!

In fact, by solving equation (3) with respect ΔC and substituting the respective CMRR values at 10MHz for both ranges it will be possible to have a preliminary estimate of the C misalignment:

In my specific case (CMRR = -39dB @ x10 range and -22dB @ x100 range), ΔC will be:

         2.15 pF   x10 range
         3.80 pF   x100 range

Considering that in the real world we are dealing with a non-ideal op-amp, the pair of values just found will represent a slight overestimation of the actual ΔC values.
In practice, in our scenario, we will have that:

         ΔC < 2.15 pF      for the x10 range
         ΔC < 3.80 pF      for the x100 range

The ΔC difference between the two ranges (about 1.65 pF), as already mentioned, is due to a colossal design error that I leave to you the pleasure to identify!!!
Without this error, the cmrr of the x100 range, between 1 and 10 MHz would have benefited by approximately 10dB!

Of course we will go further…
Instead of correcting the error we will proceed directly to the full compensation of both ranges, obtaining even better results!!


Happy WE and see u soon

ExaLab:
As explained, the anomalous CMRR degradation is mainly due to the cap. imbalance between the two op-amp inputs.
Going into more detail, the positive input has a basic excess of capacitance most likely due to the intrinsic input impedance misalignment of the used op-amp (data sheet highlights a ΔC of about 1pF)

When the x100 range is selected, a furter contribution (of the same sign…) is added to this basic misalignment. This additional capacitance is due to the incorrect circuit positioning of the relevant calibration trimmer (Fig. 1a). As it is easy to see, this trimmer represents a clear element of asymmetry between the two input branches (positive and negative).
The mistake made by Micsig (one of many mistakes...) was to place this trimmer downstream of the 1Kohm input resistor instead of directly at the buffer output. By doing so, they accentuated the capacitance present on the positive input of the differential stage, increasing the unbalance and therefore drastically worsening the CMRR of the x100 range.
Never seen such serious errors!!!

The x100 range is therefore doubly penalized: a greater capacity imbalance on the one hand, a higher Rp value on the other.
As mentioned, we will not directly fix this error… We'll balance it (more effective choice!)


Two are the solutions that I propose (alternatives to each other):


Basic Upgrade

A simple and ready-to-use solution that can satisfy most of you. This version, in principle, does not even require recalibration of the probe (which however I highly recommend...)

Required components:

            1.2 pF 1206 NPO Capacitor      (x1)
            1.5 pF 1206 NPO Capacitor      (x1)
            0 ohm  1206 Resistor              (x1)


Advanced Upgrade

This is the solution that guarantees the best results. Its hardware implementation is as simple as the basic version.
What makes it a little more complex are two aspects:
   1) To obtain the best results it requires a fairly complex calibration procedure
   2) The capacitive ceramic trimmers used may not be easily available

Required components:

            Miniature SMD Ceramic Trimmer Capacitor
            Min. cap range  0.9 pF – 2.5 pF         (x2)
            6.8 pF 0805 NPO Capacitor               (x1)


Notes:

-  For the Advanced Upgrade, as trimmer capacitors I used the Murata TZC3 Series 1.4 – 3 pF (PN: TZC3Z030AA01).
   The minimum guaranteed C is 1.4 pF.  Measuring the component, all my samples exhibited a minimum capacitance of less than 0.90 pF

-  The Advanced Upgrade requires an additional hole in the metal screen, aligned with the position of VC6

-  The proposed solutions use two auxiliary balancing capacitors: Ca5 (VC5) acts on both ranges (x10 and x100), Ca6 (VC6) affects only the x100 range

-  To open the probe, gently lift both front labels (the main and the top one). By operating delicately they will not be damaged.
   If part of the adhesive layer is damaged, remove it completely. You can easily replace it with double-sided tape


            
Next time I will describe the simplified and the full calibration procedure.
Stay tuned!

Martin72:
Excellent work so far, very great. :-+

ExaLab:
Preliminary considerations regarding calibration:

1) This probe does not provide either offset adjustments (carried out via SW) or gain adjustments (the accuracy of which depends only on the precision of the resistors used)

2) As anticipated, each of the two branches (pos and neg) of the input attenuator stage is equipped with an high frequency compensation that acts near 15MHz allowing small gain adjustments (+/- 0.1dB max).
This is done splitting the low side RC in the parallel of two distinct RC bipoles with asymmetric capacitances. The zero of the full TF that would be obtained with the single bipole solution is thus converted into a pole at the same frequency plus two zeros, one to the right and one to the left of this one. Increasing or reducing the cap. difference between the two RC bipoles is possible to increase or reduce the HF gain (note that, in order to maintain the LF Compensation, the sum of these two capacitances must remain constant!)
As I’ll explain later, this HF compensation has several drawbacks.

3) Although the calibration should be carried out with the probe in its case (by removing the central label on the back of the case), for reasons of practicality and better access to all the adjustment trimmers we will do it removing the back cover.
Once successfully carried out calibration in these conditions, it will always be possible to refine the three main adjustments (LF Compensation, DC CMRR and AC CMRR) with the device in its case.

4) The procedure must be carried out with the metal shield rigorously fixed with the two screws provided. Pay attention to the fact that for frequencies higher than a few kHz, given the low capacities in use on the HV divider, the behavior of the input front-end is heavily influenced by the position of this shield and by any minimal variation in pressure on it. So, during the entire procedure, avoid to touch it!
Later I will propose how to fix it in a stable and definitive way.

5) DC CMRR adjustment requires the use of a high input voltage. If you do not have a generator with these characteristics (i.e. >100Vac), it is common practice to use the Live terminal of the mains voltage. I assume that whoever carries out this operation is adequately informed about the possible risks involved. All safety rules and guidelines for elevated voltage measurement should be followed!

Personally, to avoid damages to the equipment in the event of a fault or accidental short circuit I connect to the “Live” of the mains voltage using a 100KΩ 1W decoupling resistor, placed inside a common power plug.

6) AC/HF CMRR should be adjusted (or evaluated) using a sinusoidal source of at least 100Vpp. Not having amplitudes of this magnitude, all available resources on the oscilloscope must be used to remove all forms of noise extraneous to the stimulus signal. I recommend to trigger the scope on a dedicated channel connected to the signal generator and then use the most suitable method to remove the noise (averaging, BW limit, hires filters and so on...).

For the less experienced, I remind you that the CMRR of the probe can be easily calculated starting from the output/input measured ratio (in dBs) by adding 20dB for the x10 range or 40dB for the x100 range (e.g. if the probe range is x100 and the output/input measured ratio is -80dB, the CMRR will be: -80dB +40dB = -40dB).

7) Before to start with the procedure, connect the device to the supply and wait at least 15 minutes



Simplified Calibration Procedure


a) LF Compensation

  1- Connect the DP10007 output to the oscilloscope

  2- Select the x10 range

  3- Connect the DP10007 inputs to a 10KHz / 20Vpp square-wave source.
      Red lead to the signal and black lead to the ground

  4- Adjust the oscilloscope to properly display the waveform

  5- Adjust VC4 in order to optimize the the square-wave response (act as if you were calibrating a common x10 passive oscilloscope probe)

  6- Connect Red lead to the ground and black lead to the signal

  7- Adjust VC2 for the best square-wave response


b) DC CMRR Adjustment

  1- Connect the DP10007 output to the oscilloscope (make sure that the oscilloscope is regularly connected to ground)

  2- Select the x10 range

  3- Connect both the probe input leads (red and black) to a ground referred 50/60Hz / 200 – 240 Vrms sinusoidal source. To do this you can use the live terminal of the mains voltage
      (Warning: use all the necessary safety precautions above mentioned)

  4- Adjust the oscilloscope to properly display the waveform

  5- Adjust VR1 to minimize the amplitude of the displayed waveform

  6- Select the x100 range

  7- Adjust VR2 to minimize the amplitude of the displayed waveform


c) AC CMRR Adjustment (LF Channel Alignment)

  1- Connect the DP10007 output to the oscilloscope

  2- Select the x10 range

  3- Twist  the probe input leads and connect both (red and black) to a 20KHz,  sine wave signal source (minumum amplitude: 20 Vpp / 200 – 300 Vpp recommended).
      Make a direct ground connection between the signal source and the oscilloscope (this can simply be done with the probe used to monitor the signal of the source)

  4- Adjust the oscilloscope to properly display the waveform

  5- Adjust VC4 to minimize the amplitude of the displayed waveform


d) HF CMRR Adjustment (for Advanced Upgrade Only)

  1- Connect the DP10007 output to the oscilloscope

  2- Select the x10 range

  3- Twist  the probe input leads and connect both (red and black) to a 10MHz,  sine wave signal source (minumum amplitude: 10 Vpp / 20 Vpp or more recommended).
      Make a direct ground connection between the signal source and the oscilloscope

  4- Adjust the oscilloscope to properly display the waveform

  5- Adjust VC5 to minimize the amplitude of the displayed waveform

  6- Select the x100 range

  7- Adjust the oscilloscope to properly display the waveform

  8- Adjust VC6 to minimize the amplitude of the displayed waveform



Clarifications / notes:

  1) All the changes proposed so far have the fundamental aim of improving (even substantially) the CMRR for frequencies above 100 kHz.
      With reference to this parameter, my probe is now able to compete with products of the best manufacturers (-60dB @ 1MHz, -55dB @ 10MHz and -35dB @ 50MHz)

  2) All the modifications proposed so far do not solve the serious problem of resonance that you may encounter at low frequencies (see the graph below)
      This anomaly, if present, involves the abrupt degradation of the CMRR for frequencies around 6700 Hz and is due to a completely different cause

  3) Given the numerous necessary premises, I must postpone the Complete Calibration Procedure to my next post

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod