On 3/19/2017 1:07 PM, Michael Thompson via cctech wrote:
>> Date: Sat, 18 Mar 2017 21:46:21 -0500
>> From: Jay Jaeger <cu...@charter.net>
>> Subject: Re: LINCtape/DECtape Head Alignment
>> Curious:  How are you measuring the signal from the head?  Do you have
>> an honest to gosh differential probe, or are you using some other
>> technique?  (If you have a differential probe, then the TU56 manual
>> indicates that you should see 10mv-12mv (the addition of the two paired
>> heads together), so as a first guess I am guessing you are looking at
>> the coils one at a time.
> We used the procedure in the TU56 maintenance manual, and used two G888
> modules to make the equivalent of the G500 described in the manual. The
> G888 modules really cleanup the high frequency noise mixed with the head
> signals.

Actually, I finally found a way to measure the heads and do it
repeatably.  Fortunately, my scope can go down to 500uv/division
(5mv/division with the 10x probes).

First, updating to the latest software on my DS2072A helped - but just a
little.  Getting the settings just right - not all of them immediately
intuitive - helped more.

(BTW, if you do not have a DSO, they are WONDERFUL.  If you haven't seen
one of these things in action, go to the EEVBLOG on YouTube My normal
analog Tek scope now has a tear in its eye.  Poor thing - it served me
well.  I got my scope during a time when Rigol had a promotional
offering, so I got the decode features and a few other things that one
might normally pay extra for.  I got my from Saelig and they offered a
discount to match that offered by other vendors who join the EEVBlog


Here is what I learned over the past couple of days:

1.  I set the bandwidth limit on the input to 20MHz.  So this limits the
bandwidth before it even gets to the Math processing and reduces the
sample rate.  Maybe this gives the 'scope more time to do other
processing?  Anyway, this made a VAST difference.  I can still see the
waveform without doing this, but it is attenuated and more quantized.

2.  At times I did not have the 10x multiplier on the input channels set
to match the probes, which meant my signal readouts were 10x off.  This
was another important mistake that lead to incorrect readings.  (D'oh)

3.  I then used the "Math" function on the scope and select a low
bandwidth filter, and set it to 50Khz (more or less matching the TU56
manual's 60Khz identification of the Tek module they used).  This gets
rid of almost all of the noise (I live less than 2 miles from a bunch of
TV and radio towers!)

This way, I reliably get decent signal readings off of the heads.

In fact, in this configuration, I could even set the Math to A-B and
sometimes see the wave as the manual offers intended, though it was
chock full of noise (unfortunately and understandably, the scope can't
do both low filter and A-B at the same time.)

(It might even be that I could use the scope to capture the data, and
then post process it, but what I have now is just fine.

It was helpful that I had reasonable confidence in the head to begin with.


Reply via email to