Topics

7904A readout bounce


NigelP
 

I have a feeling this has come up before..... I notice my 7904A has vertical readout bounce especially with low horizontal scan speeds. Any suggestions as to where I should be poking around first before I start investigating?


Roger Evans
 

'Adjust Thermal Compensations' in the vertical section of the Adjustment and performance Check.

I have recently done the same adjustmentson a 7934 and the readout is now very stable.

Roger


NigelP
 

On further investigation this is vertical bounce due to ALT mode being invoked either on the plug-in or the mainframe when lowish scan rates are used. I didn't see anything about this matter in the manual; I wonder if there is a supply decoupling problem somewhere?

Regards

Nigel


 

Thermal offsets are literally that, offset voltage drift in the
differential pairs through the horizontal and vertical signal paths.
They are caused by the changes in operating points with signal level
which result in a change in power dissipation and temperature in the
individual transistors.

At higher frequencies, the changes in power dissipation average out
but alternate mode occurs at a frequency determined by the sweep
lengths so it is slower than any individual sweep and slower yet than
the vertical signal.

If you want to see the effect in a controlled way, watch how the
readout shifts when X-Y mode is used and the beam is moved to
different points on the display. This is (usually) not the result of
electrical leakage between the beam position and readout; it is caused
by changing the operating point of the transistors ultimately changing
their temperature which changes their current gain and base-emitter
voltage.

On Thu, 14 Dec 2017 21:56:26 +0000, you wrote:

On further investigation this is vertical bounce due to ALT mode being invoked either on the plug-in or the mainframe when lowish scan rates are used. I didn't see anything about this matter in the manual; I wonder if there is a supply decoupling problem somewhere?

Regards

Nigel


NigelP
 

Seems odd, but let's be quite specific about what I see; the readout jumps up and then down exactly at the start of each alternate sweep; difficult to imagine that this is a thermal problem... it's almost as if a switch is being thrown one way and the other at the start of each sweep (and indeed somewhere in the sweep circuitry that is probably exactly what is happening.... by design)? Also, the effect is quite flip-flop, so not increasing with sweep position but actually at the start of each sweep; one might argue that it's coincident with each sweep gate.

What I have not yet checked is whether fully overlaying two or four alternate traces on the screen minimises the effect; in other words whether it is the shift in vertical displacement of the normal two or four scan lines that is attributing to the readout display shift.

Anyway the bottom line is that the thermal compensation adjustment process is not actually solving the issue. I can see that if I adjust (eg R131 @ 50mS sweep) to one extreme or the other I can observe readout jitter going in or out but it doesn't seem to minimise the bounce.

Nigel


Roger Evans
 

I have just had a look to see what my 7934 (same Y amplifier and channel switch) does at slow sweep speeds. If I put one of the Y traces near the top of the screen and one at the bottom and set the sweep speed to 100msec/div then I see a small shift of the top and bottom readout positions as the beam alternates between the two traces. The movement is at most 0.05 of a major division, quarter of a minor division When the upper beam is displayed both readouts move slightly downwards, when the lower beam is displayed the readouts move up slightly. If I put the two traces near each other, anywhere on the screen, the readout does not move. At 10msec/div the movement is about the same but more objectionable, at anything faster than 1msec/div the readouts are steady. This seems consistent with thermal effects and David explained it very clearly.

IWhat sweep speeds are you using and how much movement of the readout do you see?

Roger


bobh@joba.com
 

David,

I haven't tried to look at this effect (moving the beam position) and
its relation to the readouts.� Does it depend on which scope you are
using and the range selected.� Seems like higher ranges (if available)
could have more affect on the operating point.� I would have thought the
diff amps would be tolerant of this kind of effect.

Your analyses are usually a couple grade levels above me but I would
like to understand in order to apply to troubleshooting. And, it seemed
like potential for a nice on-topic discussion.

Bob.


On 12/14/2017 6:18 PM, David @DWH [TekScopes] wrote:

Thermal offsets are literally that, offset voltage drift in the
differential pairs through the horizontal and vertical signal paths.
They are caused by the changes in operating points with signal level
which result in a change in power dissipation and temperature in the
individual transistors.

At higher frequencies, the changes in power dissipation average out
but alternate mode occurs at a frequency determined by the sweep
lengths so it is slower than any individual sweep and slower yet than
the vertical signal.

If you want to see the effect in a controlled way, watch how the
readout shifts when X-Y mode is used and the beam is moved to
different points on the display. This is (usually) not the result of
electrical leakage between the beam position and readout; it is caused
by changing the operating point of the transistors ultimately changing
their temperature which changes their current gain and base-emitter
voltage.

On Thu, 14 Dec 2017 21:56:26 +0000, you wrote:

On further investigation this is vertical bounce due to ALT mode
being invoked either on the plug-in or the mainframe when lowish scan
rates are used. I didn't see anything about this matter in the manual;
I wonder if there is a supply decoupling problem somewhere?

Regards

Nigel


[Non-text portions of this message have been removed]


NigelP
 

So, further investigation demonstrates more of what I've already noted. Scenario: 7904A loaded with 2x 7A26, 7B92A, 7D15 and ALT set on both vertical PIs and mainframe set to ALT...... if I overlay all four traces on top of each other there is NO bouncing of the readout . As you increment any trace position towards top/bottom of the screen the bounce gets increasingly larger reaching around 0.1 major division (therefore readout bounce magnitude is a direct function of trace vertical position). Sweep speed in the region of 50mS/Div +/- some shows up nicely, but any variation of sweep speed simply varies the level of visual annoyance. Any combination of ALT setting produces the same result.

Similarly, substituting a 0587 Signal Standardiser into either PI slot and set up to perform the handbook "Thermal Comp" procedure, as mooted earlier in the thread, the readout bounces away merrily to the step waveform rising/falling.

Also using for example a single 7A26 set to CH1 and mainframe set to LEFT (so no ALT settings) the readout shifts up and down according to PI vertical position setting.... a direct and linear relationship between SHIFT potentiometer setting and readout vertical screen position. Readout movement is in anti-phase to trace shift movement.


cmjones01
 

On Tue, Dec 19, 2017 at 5:19 PM, NigelP <nigel-pritchard@...> wrote:
So, further investigation demonstrates more of what I've already noted. Scenario: 7904A
loaded with 2x 7A26, 7B92A, 7D15 and ALT set on both vertical PIs and mainframe set to
ALT...... if I overlay all four traces on top of each other there is NO bouncing of the readout .
As you increment any trace position towards top/bottom of the screen the bounce gets
increasingly larger reaching around 0.1 major division (therefore readout bounce magnitude
is a direct function of trace vertical position). Sweep speed in the region of 50mS/Div +/-
some shows up nicely, but any variation of sweep speed simply varies the level of visual
annoyance. Any combination of ALT setting produces the same result.
I've just tried this on my 7904A and I also get some readout bounce,
but very little. If I position two traces in ALT mode, one at the top
and one at the bottom of the screen, so they're only just not visible,
the readout bounces a very tiny amount (timebase at 10ms/div), much
less than 0.05 divisions. You really have to look carefully to see it.
If I crank the Y position knobs to their maximum possible extents, I
can get the bounce up to about 0.1 divisions, but that's the extreme.

This 7904A hasn't been refurbished other than a slight tweak to the Y
gain and to one of the thermal compensation adjustments. I use it in
my work almost every day and the readings it gives appear to be
accurate when compared with other instruments I use. It doesn't have
traceable calibration but I'd say it was within spec.

Chris


NigelP
 

OK, I think this is now the final input from my side on this topic!

I studied the schematic and noted something that I thought odd in comparison to the test schedule F5. I can see that the pot R130 is said to adjust the 1MHz jitter with 1uS/Div. Now, without a detailed analysis it seems to me that R130 is adjusting a "DC" or LF component since it does not have the capacitor associated with all the other compensation pots (except for C200 connected around the op-amp). All other adjustment pots are associated with a capacitor value which tracks the rep-rate value you are trying to compensate (higher value for lower frequency and vice versa). Admittedly all pots do have some level of DC adjustment but R130 circuitry seems to be a bit more DC-aggressive (maybe just my weird way of thinking about things :().

Now when I first tried this F5 procedure, noting that 1MHz did not need any adjustment (there was no jitter) I didn't bother to even try adjusting it, and just concentrated on R132 & R131 which is targeted at 10Hz which is typically where all the problem exhibits itself.

I decided to try out my theory that R130 ought really to be responsible for DC/LF jitter/bouncing correction; and indeed it does! I was able to tune out the 10/100Hz bounce far better than with the supposed correct pots R132/131 :). Re-evaluating the 1MHz rep rate results showed no loss of readout jitter compensation, and indeed I did not even bother to check the C200 adjustment. So simply speaking, first adjust R130 at 10Hz and then follow the procedure as defined in Table5-5.

So I now have a quite satisfactorily compensated readout across the full range of test offered by the 067-0587-01 and 067-0587-10 Signal Standardisers. I also found that using the Gain setting on the Standardiser (a full screen height staircase waveform) rather than the Step Response setting gave a far more aggressive signal for the compensation to be adjusted against; it was more visible and in effect the rep rate of the test signal is then one-eighth of the Step test. I can just observe a marginal vertical movement but nothing that is going to bother me.

I did also notice that adjusting R130 against 10/100Hz did actually vary the overall top-to-bottom magnitude of the readout display versus the Y-channel waveform; perhaps to be expected?

So reviewing the F5 test schedule and Table 5-5, firstly (in my manual at least) there appears to be a typo error...... step-5 indicates a rep rate of 100KHz (which is actually step-2) and should read 100Hz. Also in my personal experience the first step should be to adjust R130 at 10Hz and then check at 1MHz (perhaps adjusting only C200 where necessary) before moving on to the other steps in the table.

Comments on the above discoveries from those more knowledgeable/experienced are surely welcomed :).

Nigel


cmjones01
 

On Wed, Dec 20, 2017 at 1:48 PM, NigelP <nigel-pritchard@...> wrote:
I decided to try out my theory that R130 ought really to be responsible for DC/LF jitter/bouncing correction; and indeed it does! I was able to tune out the 10/100Hz bounce far better than with the supposed correct pots R132/131 :). Re-evaluating the 1MHz rep rate results showed no loss of readout jitter compensation, and indeed I did not even bother to check the C200 adjustment. So simply speaking, first adjust R130 at 10Hz and then follow the procedure as defined in Table5-5.
That sounds great - well done for tracking this down, and thank you
for documenting it. I should look at the adjustments on my scope and
see if I can make it even better than it already is.

Chris