Calibration problem with a DSO3202A

[mwbarth36]

 I posted this in Jan.  Still working on the scope.   


{{I have a DSO3202A scope which has the following problem, it will not run the 'Self Calibration' sequence.  Continues to display the message "Disconnect all Inputs", even when all inputs are disconnected.  What's more, looking at the inputs there is no sign of abnormal voltage or resistance. I suspect the problem is with channel 1.  That is where the Self Cal. program starts, and it never advances past trying to get channel 1 input to indicate that there is nothing connected.  Also, both channels operate normally with signal, and are right on, spec. wise.  The only thing I can find is that when Channel 1 is switched from the 100 mV/div position to the 50 mVdiv position the reference is no longer 0, but goes to a -150mV level, and stays at that level when switching lower to the 20 mV/div position.  All other positions of the vertical settings are 'spot on', and the scope appears to operate normally, with AC gain accurate on both channels.  Normally I would expect the Self Cal. to correct problem at the lower setting on Ch. 1.......except the Self Cal. will not start because the program 'thinks' there is something connected to one of the inputs, presumably Ch 1.

Of course, it appears impossible to get a circuit diagram for this scope.  If wrong please let me know.  I would appreciate any help I can get in solving this problem, as the scope appears in good shape other than this problem.  I have opened the scope and compared measurements of components in the channel 1 and channel 2 attenuators, and can find nothing different between the two.  Cannot power up the scope in the disassembled state to check voltages.

Any ideas greatly appreciated!}}

I now have an operational 3202A that does calibration just fine.  Of course it is not possible to get schematics for this scope, and does not appear that I can operate it when disassembled.  Here is what I have learned.  Almost certainly, the Calibration sequence is starting OK only to sense that there is still something attached to an input, probably channel 1.  There is not, the impedance is 1 Meg. and no voltage of any kind.  When the cal. sequence starts there is a relay activated, a white square appears in position 1 on the display, then a bit later another relay fires, or tries to, and the message 'disconnect all inputs' appears, or remains, on the display, not sure which.  The sound of the relays firing does sound a bit different in the operating scope and the bad one, not sure what that means.  It is my guess that a relay is not connecting the cal. signals to the channel input.  Does that sound like a valid assumption, anyone?  Has anyone out there experienced a similar problem.  Any help, or suggestions will be appreciated!

[Renate]

I found this interesting because just today I was working on writing my own calibration routines for the Hantek 6052BE USB oscilloscope.
Ok, I know that it doesn't direcltly correlate with your problem.

On my scope a DAC provides bias for zero adjustment and trace positioning.
For me to calibrate I have to find the DAC value that puts ground in the center of the ADC converter range for each of the 9 gain ranges and each channel
I then have to determine the numeric gain factor for the DAC that allows me to position the trace where I will.
There is only one factor for each channel.
I'm not sure at what price range you get an actual channel gain calibration on an oscilloscope.

How many relays does each channel have?
Mine has one for AC/DC coupling, and two for x10 attentuators.
Unless you have more then there is probably no way to feed in an actual calibrate signal.

In any case, on your bad range, you are off by 3 divisions, that's 3/8 of the DAC range, a pretty serious error.
Of course a bad calibration could easily set it that far off.
Is there any way to read the calibration data off the scope?
Is there any way to write the calibration data?

Have you tried grounding the input to see if changes the offset?

[Renate]

Just to continue that thought...

On my scope the ADC is 8 bits, the DAC is 12 bits.
The (reverse) gain factor on the DAC to ADC is 6.14
That means that the DAC has to increase by 6.14 LSB to raise the ADC by one LSB.
Since the DAC has 4 more bits that means that the DAC full-scale range is 16/6.14 times the ADC full scale range.
Let's call that 2.6 times.
So there is plenty of range in the DAC adjustment.
It means that the scope could display DC coupled signals full window with the zero line below the window.
For example, if a wave was +1.3V positive peak and +0.3V negative peak it could display it as a full window on a scale with 1V FS.
The zero line would be off scale.
I know of no scope that has the user controls to have the zero line below/above the window.

In any case, the DAC itself has a low pass filter on it to prevent injecting noise into the channel.
The time constant might be as low as one second.
This has an effect on how to calibrate.
Trying to zero calibrate we might naively think to do a binary search for the correct value.
That would be trying a value, then seeing if the offset was plus or minus.
Like: 0x0800, 0x0400, 0x0600, 0x0500, 0x0580....
Unfortunately the crazy swings at the start would not work out well with the low pass filter and the long settling times.
Instead we start with what we presume is a half decent previous value and make tiny adjustments from there.
Like: 0x07ed, 0x07ec, 0x07eb...
If there were a previous error or a major change in the offset a (stupid) calibration routine might give up before it found the correct value.

So where is your offset coming from?
1) Bad calibration value and the calibration routine is too stupid to find the correct value if it starts at the wrong point.
2) DC leaking into some part of the attentuator stage? Contaminated flux between a relay coil pad and contact pad? Contaminated relay? Since it's going negative, the leak would have to be negative.
3) If you actually have a path for injecting an actual calibration square wave into your channel, maybe there?

I think that I'm going with #1.

[mwbarth36]

First, let you thank you for your response.  What you are attempting is something I would not attempt, as my background is analog circuit design (have been retired from Engineering for 16 years), still, I am fascinated reading your report on what you are attempting.

I do believe the situation I have is a bit simpler, i.e. the offset is correctable, I believe, but the scope cannot run the calibration program because it does not get past the first step. I do believe it is a relay that is not being activated, one that deals with only the calibration phase of operation.  I believe this because the scope runs through every vertical setting accurately when signal is applied.  Of course, without a circuit diagram of any kind available, I can only guess at the circuit.  I have accessed the atten. section of the circuit, there are 8 relays in total, 6 are small double pole, double throw, and 2 are reed relays.  All coils  appear to be intact (measure appropriate resistance).  Cannot tell which one, or ones, deal with the calibration phase.  I suspect one of the reeds, as they a prone to failure by simply not closing.  Remember, this scope starts the Cal. phase normally, a relay fires, the message 'disconnect all inputs' disappears from the screen, and a white square appears in the first position of the Cal. Bar on screen.  Then, another relay should fire to continue the Cal. (this is what happens on a good scope), but on the bad one no relay can be heard, and the message to 'remove all inputs' reappears on the screen. 

Let me know what you think of my analysis, or please, give me your thoughts.  I believe I have set the scope up to where I can access the attn. section with the scope operational, but have no idea what to look for other than to try to see if any relay driver (these appear  to be transistors) is trying to drive one of the relays at the time the during the Cal. sequence when the second relay should fire.

Again, any thoughts are appreciated!

[Renate]

Without a schematic it's hard to tell.
If you have a good photo of both sides of the PCB around the relays we might be able to tell.

Here's an experiment:
Put the scope on the range where it's 150 mV lower than it should be.
Connect a positive power supply/battery resistors/pot to the input to get rid of half of that difference.
Don't make it compensate the whole difference.
Hit calibrate and see if you can fake it into making a run-through.
Remove your input.
Has the 150 mV error been reduced (maybe to half)?
Can you get calibrate to run now with no input?

[mwbarth36]

Well, finally did get this scope fixed.  One would never guess the problem.  While looking over the attenuator section for channel one, I noticed a resistor that was slightly burned looking just to the left of the attn.  Could not read what the value was suppose to be, measured with an Ohm meter and it was obviously open.  Since I have a second functioning 3202A, took a look at what the value should be (249 Ohms).  Then noticed the OP27 in the bad scope was installed backwards from the one in the good scope.  This OP Amp is used to generate the calibration signals from the DAC and apply them to the input.  I do believe this OP Amp was installed wrong at time of manufacture.  Anyway, when one installs this backward in this circuit, the +V supply is connected to the + input pin and the output since the Amp is used as a 'Follower' with the + input pin connected to the output.  What a mess that caused.  Took out a resistor in the atten., three resistors associated with the op amp, and the 74HC4053 sending the DAC signal to the op amp.  Replaced them all and the scope now runs calibration perfectly, and left with zero offset.