开云体育

Hello All,
Following the restoration of my last acquisition, a Tek 7623A, after fixing a few minor issues (Fan, Compressed & Inverted Horizontal sweep, MIX time base switch broken, etc...), and after consulting with the group on best approaches to calibrating the Vertical Amplifier without having a Signal Standardizer plugin, I came to the point of calibrating the Vertical Amplifier of this lovely beast.
Following some of the group's advice, I already manage to get the DC calibration department right (Gain - R2446), and also, the Thermal Balance calibration (R2527).
Getting to get a decent square wave displayed on-screen, however, always puts up some challenge and, as frightened as I always am about the highest frequency performance, it seems that this chapter is (has been) much more a challenge for the designers than it actually is to the person (me) doing the calibration...In other words... Clean fast edges and sharp flat-corners around the very first few nanoseconds always seem to be more easily obtained that I anticipate.
However, I always seem to have problems to get those damn medium to long-term flat-tops, flat.
Therefore, the main issue of this topic.
After the initial fast edge reach 96% of the full DC level (4% lacking), the square wave top takes about 3us to reach 100%.
I made 2 YT videos showing the present situation and the insufficient effect of the only two adjustments that have some influence, R2409 (1st microsecond or so) and R2509 (1st 500ns or so).]




I checked already.
1. Power Supplies Voltage and Ripple / Spike OK: +50V, +50 DCPL, +15V, +15V DCPL, -15V and -12.5V.
2. Decoupling capacitors OK: +50V DCPL (C2593 0.1u), +15 DCPL (C2595 15u), -12.5V (C2597 15u), 2nd Stage / Thermal Balance Bias (C2524 1nF), 1st stage / Dly Line Bias (C2403 10nF)
3. Voltages and waveforms at amplifier's output on the right P-P range (for the displayed waveform) and DC levels (about +25V).
4. DC Voltages at different biasing points OK.
5. Plugin output waveform (at pins A11 and B11 of backplane connectors) don't display this 4% lacking during the first 3us.

6. Maybe a problem here... Signal coming out of delay line, entering at J2401 / J2501, seem to show more or less the same 4% lacking, slowly getting to 100% in about the same time... but it's more like a normal, rounded square wave.
It doesn't display a clear knee at about 96%, and it looks more like a square wave lightly integrated (i.e.: asking for some HF boost).

7. I fully removed the Vertical Channel Switch Board, cleaned contacts, removed and reseated the Tek custom vertical switch (U2340) and it didn't show any improvement.

8. Last, but not least, while checking the waveforms at the bases and emitters of Q2441 and Q2541, the looked rather different (the peak-peak amplitude at one of them was much bigger than the other (about double)...
However, there's no distinguishable difference of the waveforms at their collectors (maybe because they are balancing out?)
Taking both transistors out, and testing them on my AVR transistor tester, one showed distinctively different than the other, more than I would consider normal for 2 transistors being used on a complementary stage like this.
A difference of almost 20mV in Vbe (one have 700mV the other 720mV), and a difference of about 30% of hFE (one have 35, the other almost 50).

Questions:
A. Any ideas... suggestions on further checks / tests to do?
B. Can those transistors be causing the problem? What would be a modern, currently available equivalent to an A5T4261 transistor?
It's a 2GHz fT; 2.5pF Ccb; >30hFE; 12V VCEmax, 80mA Icmax

Thank you all,
Rgrds,
Fabio

Hi Fabio,

Are you sure this problem is in the mainframe Y amplifier and not the vertical plugin? I assume that if you don't have a standardizer you have to go via a normal Y amplifier. What are you using?

Regards,

Roger

Roger,
Well... I can't tell I`m 100% sure.
I can tap with a x10 oscilloscope probe onto the 11A and 11B signals coming out of the 7A26 Vertical plugin, and the waveform, although not perfectly square (it's a little roundy in the corner), it doesn't show that slowly flattening voltage over this long 3us time period...
I assume the non-perfect corner I see (while probing the plugin output) is because I cannot properly reference the probe with a short ground lead so, I cannot take seriously anything immediately following the sharp edge. but the passive probing + long ground lead could not be concealing this 3us long voltage lagging.
If the problem I would be having would be in the first 100ns or so, I reckon that any artifacts I see could be coming either from the plugin or from the mainframe misalignment (or a combination of both), but 3us is just too much.
For a circuitry of that grade... anything slower than 1us is almost the same as DC... lol.
I also reckon that in the 10s or 100s of milliseconds, there can be other disturbances at play, like thermal imbalance (between different sides of complementary designs) or power supply mains ripple... but that's not the case here.
Yesterday, while I still didn't have chance to probe the P.S. voltages (inside the board, after decoupled), I was really convinced there was a decoupling issue... that right after the fast corner which consumes a lot of current... the available voltage was depleting and the output would be - then - being limited by the available supply, until the - lacking - decoupling would have time to recover.
But yesterday evening I probed all the important voltage supplies and bias points, and they're pretty much rock steady.
Not happy, I added additional decoupling in parallel, to each of those voltages... and neither made any difference to the displayed waveform so, I think it's safe to conclude that there's no decoupling issue (at least not on the Vertical Amplifier board).
I still need to find ways to rule-out a possible decoupling issue on the Vertical Channel switching board, that's difficult to reach and to probe at.

Happy easter to you all.

FT

Fabio,

I have just checked on my 7A26 and the Low Frequency Compensation (attenuator) adjustments do have about the correct time constant to correct for your 3usec lag in response (I only checked one of the attenuator settings, 0.1V and C114,C115). You could try using the 7A26 LF compensation to flatten the 1 - 3usec region before going back to the mainframe adjustments for the higher frequency components. I am sure you can see why the standardizer plugin is useful - you have HF compensation both in the mainframe and the plugins.

I think the sharp 'knee' you have shown on the YT video is due to trying to correct the low frequency error in the plugin with the HF adjustments in the mainframe. Adjusting the lowest frequency / latest time errors first reduces the interaction between the adjustments.

I hope I am helping rather than adding confusion, best wishes,

Roger

On Thu, Mar 29, 2018 at 01:26 pm, Fabio Trevisan wrote:

8. Last, but not least, while checking the waveforms at the bases and emitters
of Q2441 and Q2541, the looked rather different (the peak-peak amplitude at
one of them was much bigger than the other (about double)...
However, there's no distinguishable difference of the waveforms at their
collectors (maybe because they are balancing out?)
Taking both transistors out, and testing them on my AVR transistor tester, one
showed distinctively different than the other, more than I would consider
normal for 2 transistors being used on a complementary stage like this.
A difference of almost 20mV in Vbe (one have 700mV the other 720mV), and a
difference of about 30% of hFE (one have 35, the other almost 50).

Questions:
B. Can those transistors be causing the problem? What would be a modern,
currently available equivalent to an A5T4261 transistor?
It's a 2GHz fT; 2.5pF Ccb; >30hFE; 12V VCEmax, 80mA Icmax

Fabio, 151-0271-00 is not a matched pair, so Tek didn't judge that necessary . 20 mV difference is not important at all since these transistors are current driven at the emitter. The transistors in the previous pair Q2405/2505 have (or should have) enough headroom in Vce (maybe check this). This also explains that you see no difference in collector signal. Difference in hfe could do more harm perhaps, though I also doubt this.
A suitable replacement would be 2N4261. Sphere has a few new in stock under 151-0434-00, $6 each. In the cross ref also listed as 151-0271-00.

Albert

Hello Roger,
Apologies for the long delay for feedback... I was getting my hands dirty on the thing.
See my comments right after yours...

I have just checked on my 7A26 and the Low Frequency Compensation (attenuator)
adjustments do have about the correct time constant to correct for your 3usec
lag in response (I only checked one of the attenuator settings, 0.1V and
C114,C115). You could try using the 7A26 LF compensation to flatten the 1 -
3usec region before going back to the mainframe adjustments for the higher
frequency components. I am sure you can see why the standardizer plugin is
useful - you have HF compensation both in the mainframe and the plugins.

I confess that when I read your message, my first reaction was to dismiss your suggestion, because you meant adjusting the input attenuator, and since I am using - on purpose - a 25mVpp signal, I am using the 5mV/div range, which doesn't engage any attenuator that I could compensate for.
But, your message did trigger my curiosity about what could possibly be the AC frequency compensation adjustments on the plugin and - bingo - besides the attenuators (which were ruled out), the 7A26 has not one, but two LF response adjustments, R1436+C1436 with average time constant of 150us, and R1431+C1431 with avearge time constant of 5.5us.
And not surprisingly, those 2 adjustments together, in combination with the lowest frequency response adjustment of the MF's vertical amplifier, have plenty of room for correcting the low / mid frequency lacking that I was finding.

I think the sharp 'knee' you have shown on the YT video is due to trying to
correct the low frequency error in the plugin with the HF adjustments in the
mainframe. Adjusting the lowest frequency / latest time errors first reduces
the interaction between the adjustments.

Yes, I always try to go from LF to HF, but I was kind of blinded by the fact that the signal I was seeing (at the plugin's output) was free of that LF lacking, so I was convinced that I had to pursue the solution at the MF's amplifier alone.
But the bottom line is, I don't have experience enough with the 7000 family to know what would be a correctly looking signal at the plugin's output, never mention that my test rig was less than ideal in this case, looking at only one side of the balanced output, with a long ground lead... my assessment of the "correct" signal coming out of the plugin was, at least, subjective.

I hope I am helping rather than adding confusion, best wishes,

Not at all, your suggestion pointed me in the right direction and I managed to get the overall vertical step response well adjusted.
I`m not getting (yet) a step response as fast as I managed to get from my 464 of 3.5ns... I can only get to 4ns, but I may still be getting confused about the great number of adjustments there are (between the MF and the Vertical plugin), and the almost infinite possible combinations...
But I`m happy for now... Good enough for government's work!

There are a few other issues that are bugging me (on the horizontal axis... an intermittent jitter), and I will pursue that first, before trying to perfect the vertical any further... but that's subject for another post.

Rgrds,
FT

Albert,
Thanks for the heads up... my remarks just after your quote.

On Fri, Mar 30, 2018 at 01:12 pm, Albert Otten wrote:

Fabio, 151-0271-00 is not a matched pair, so Tek didn't judge that necessary .
20 mV difference is not important at all since these transistors are current
driven at the emitter. The transistors in the previous pair Q2405/2505 have
(or should have) enough headroom in Vce (maybe check this). This also explains
that you see no difference in collector signal. Difference in hfe could do
more harm perhaps, though I also doubt this.
A suitable replacement would be 2N4261. Sphere has a few new in stock under
151-0434-00, $6 each. In the cross ref also listed as 151-0271-00.

As you will see on my other reply to Roger, I managed to make the proper vertical LF step response right, by using the LF adjustments there are on the vertical plugin.
Those transistors though, seem to have become less important (but opportunely, I will try to find a replacement).
I also found, on the Tek semiconductor x-ref, that there seem to be a difference between the part 151-0271-00 and the 151-0434-00.
While the latter is indeed a direct equivalent to the 2N4261, it seems the 151-0271-00 is a variation of the 4261, which they called the A5T-4261.
The a5T kind is an 80mA Ic part, while the vanilla 2n4261 is a 30mA part.
The A5T has a TO-92 package and seems also to be capable of dissipating more power as well (0.5W against 0.2W the TO-72A package of the 2n4261).

Krgrds,

Fabio