Near The Beginning of an era, The Tektronix 511A

The acquisition of this model brings this collection and valve era narrative very close to the beginning of (what I say is) the golden era of valve electronics. Introduced in 1950, the model I have is most likely from that year; the manual was revised completely at serial number 2869 in 1951 while the example in this collection is serial number 1398.
Here it is displaying 100μS time markers:

The model 535 was introduced just 4 years later in 1954, and the rate of improvement in technology was high! This scope is quite crude by comparison though it works better than the “laboratory” Heathkit IO-14 I have which is a later design.
The case has been amateurishly repainted eggshell blue, in the process destroying the light crackle texture of the original paint. I am not good with paint and I think I will have a professional re-paint the case at some point….
It has a flat-face 5CP1A CRT operating at 3kV, the brightness and focus are excellent. Including the CRT, there are 33 valves plus 11 NE-2 neons.

All it took to get it (initially, more on this later) working again was the replacement of one electrolytic that was too far gone to re-form, it had taken out a 6X4 rectifier. I did a few other things to get it working well:
The time-base coupling cap was leaky, around 50nA. Unusually, this instrument does not use “bumblebee” caps and no other paper in oil types needed replacement though the Y shift de-coupling “bathtub” capacitor is leaking oil but not current, yet. Since there was an HF buzz on the 225V rail, I also de-coupled the CRT HV supply oscillator (from the 225V rail) using a 390Ω resistor together with a 10μF cap, this completely eliminated the buzz. I don’t feel badly about this change, all subsequent Tektronix scopes had such de-coupling.

With the 511 model, Tektronix introduced the use of regulated power supplies, an essential step in the process of transforming the oscilloscope from novel but qualitative to essential and quantitative. In the 511A every supply line was regulated and the CRT HV supply was redesigned from a line-frequency transformer-rectifier design to a HF oscillator-rectifier design that brought freedom of the CRT acceleration voltage from line variations since the HF oscillator was powered from a regulated rail. This development improved the calibration stability since the sensitivity of a CRT is inversely proportional to the acceleration potential.

The Y amplifier is not DC coupled or easily calibrated. The “calibrator” puts out a user (and line variation) variable 60Hz signal from the heater supply and is not a square-wave, I can see no way to be sure what the calibrator output level actually is, it appears to be useless to me. In the revision starting with serial number 2869, this fault was addressed. Clearly Howard (Vollum) and his men had the talent and ideas but were still finding their feet at this point, even so what they were offering was ahead of that of the market leader Dumont; Alan Dumont was more interested in the burgeoning new world of television. The excellent Tektronix calibrator did not arrive until the advent of the 315 in 1953 and further improved for the 530 series scopes in 1954. The topology is all pentode having a simple long-tailed-pair deflection amplifier that is AC coupled to the CRT plates preceded by one or two stages, selectable by a switch on the front panel. The extra stage brings in an additional 20db (X10) of gain. There is a compensated 1X, 2X, 4X, 8X attenuator and a variable control having a range of more than 2:1 thus providing continuously variable sensitivity over the entire range. There is no provision to set or calibrate the Y amplifier and given the uselessness of the “calibrator” as a reference, this aspect is a far cry from what we are used to from Tektronix and completely eclipsed by the subsequent 530 series models. For this example in 1 stage mode, the bandwidth is 10MHz and the sensitivity is circa 1V/cm down to 20V/cm; in 2 stage mode, the bandwidth is 8MHz and the maximum sensitivity is circa 100mV/cm.
An odd feature that to my knowledge was not repeated, was a switch that shifts the Y amplifier bias + or – to permit the full amplitude capability of the amplifier to used to display + or – peaks. I doubt if anyone made much use of it. I was not aware of it at first and thought there was a bias issue with the Y amplifier; there was! An intentional one. Here is the Y amplifier schematic:

The timebase though triggered, does not have the sweep length limiter and hold-off features of the soon coming 530 series designs. (The sweep length limiter cuts off the sweep generator into flyback mode at a precisely adjustable level, The hold-off circuit locks out the sweep gate until sufficient time has passed for the sweep generator to reset or flyback into the wait state, this prevents premature re-triggering and makes for much less “fussy” operation.) The time base of the 511A is fussy to say the least and really confirms just what a breakthrough the trigger – gate – disconnect diode – sweep generator – length limiter – hold-off topology really was, everybody copied it! This new topology is attributed to Dick Ropiequet; known as the Ropiequet wide band sweep circuit, it was first employed in the 315 oscilloscope in 1953, then refined by John Kobbe for the 530 series oscilloscopes which arrived in 1954. Returning to the 511, what made it unique was that the time-base was triggered, not repetitive and thus could be calibrated in time/cm. I did a rough set-up and then checked it using a marker generator and without fine tuning it, the timing was better then 5%. This was excellent performance in 1950 and I can see that with some tweaking of aged components, it can do better than 5%. This makes the difficulty of calibrating the Y axis even more puzzling. The sweep rates are 10mS/cm switchable in 5 ranges up to 100nS/cm. A variable control is provided so that each range can be slowed by a factor up to 10 thus the slowest rate is 100mS/cm. The time range legends do not specify the display traverse parameter, e.g. the “cm” in mS/cm. I realised that the markings relate to the full display width, that is 10cm so the legends are actually time/10cm. The variable control is a curiosity. It has a dial that was engraved to match the characteristics of the actual potentiometer. The production test technician would mark the dial with a pencil and then send it to be engraved! Replacement pots were supplied with a new dial.

There is a very interesting sweep magnifier feature. It switches in an additional gain stage that is arranged to have a gain of 5 together with a magnifier position control that operates from the point of sweep start to the right. What this does is to cause the trace to sweep at 5X the set value and by turning the position control, it is possible to slide a feature of interest on the signal that would be out-of-view due to magnification to the right of the screen, into view. This is a little bit like variable time-base delay, however it cannot be calibrated for time so is simply a visual tool rather than an analytical or quantitative tool. Clever and fun play with though, and as I maintain, helps to make staring at an oscilloscope screens MUCH more interesting than TV!

The sweep is amplified by a long-tailed-pair deflection amplifier, similar to the Y amplifier. While the Y amplifier is AC coupled to the CRT plates, the X amplifier is DC coupled using a chain of 5 NE-2 neons for each plate. DC coupling is necessary to get good slow sweep behaviour however simply DC coupling would put the X defection plates at a higher average DC potential to that on the Y plates thereby causing de-focussing. The neon chains are used to DC level shift the amplifier plate potentials to a value close to that on the Y deflection plates. (Cossor used the same technique in the 1035 twin beam oscilloscope. I had one of those as a kid, it had the sharpest trace I have ever seen, so good that it might have been gas-focussed but was not.) Since the impedance of a neon lamp increases with frequency, each chain is AC bypassed with 1nF capacitor. The ionisation voltage of these neons increase with age and I did find it necessary to replace them. One issue that I have not resolved (yet) is that the lamp strings “bobble” around a bit causing some lateral fluctuation of the trace. Here is the X system schematic:

Rescue required, the power transformer failed! The winding associated with the failed electrolytic and 6X4 rectifier failed, not a co-incidence. I worked out a solution using a spare Tek 316 transformer that I have. Even though the 316 transformer has a smaller core, the VA requirement is close, both scopes require a 3A fuse.
Please work with me to educate eBay sellers NOT to do the “powered up, light came on, no further testing performed” routine, this KILLS vintage equipment. That the light came on is meaningless in terms of function, this so-called “testing” simply either makes for more work for restorers or results in a great piece of history being scrapped prematurely.

Many adaptations were required, the schematic of the adaptation is shown below. The general topology of the supply is similar to that of all later Tektronix valve oscilloscope power supplies, the raw supply for each rail being stacked on top of that of the next lower rail. The major difference is that full wave valve rectifiers were used with centre-tapped windings, all later scopes including the 316, used single-phase windings and bridge rectification. For this rescue, I added silicon bridge rectifiers ahead of the valve rectifiers that now act mainly as slow-start devices. It was necessary to do some juggling to obtain enough voltage margin for the negative and 225V rails. This left me a bit shy for the 450V rail, fortunately the current requirement here is small around 17mA, so I used a voltage doubler to obtain an adequate voltage margin for the 450V supply. In fact, I now had some in hand so rather than stacking the rectifier for the 450V rail on top of the rectifier for the 225V rail, at some 315V, I relocated it on the 225V rail itself. In doing all this I aimed at and succeeded in having all supplies stay in regulation down to a line voltage of 110V. Another piece of finessing I did was to relocate the screen grid of the 225V error amplifier from the output side to the input side resulting in a degree of feed-forward that both improves line regulation and reduces ripple. A 100nF bypass capacitor was found by experiment to be optimal in injecting inverse-phase ripple into the error amplifier. Ripple fell from around 40mV p-p to around 15mV p-p while line regulation, over a range of 115V to 125V improved from 1V shift to 0.3V shift. Tektronix later used the feed-forward technique in some regulators.
I drew up a table of heaters, currents and cathode potentials. The object was to group cathode potentials to ensure that all heater / cathode potential differences are within the limits of specification for each valve. Then it was necessary to group the heaters such that each heater supply is within the current available from each winding on the 316 transformer. It all worked out perfectly!
It was necessary to add a supplementary 5V/4A transformer for the two 5V4 heaters though (which takes 20VA off the smaller 316 transformer). Here is the modified power supply schematic:

To help keep my head straight while I made the necessary changes, I made a list:
Move V21 heater from separate winding feed to the heater pins on to V25.
Connect V21, V25, V15 to winding 27/28.
Disconnect V21 heater from cathode, connect V25 heater to cathode.
Disconnect V27 heater from common heater supply.
Connect V27 and V26 heaters together.
Connect V27 and V26 heaters to winding 23/24.
Disconnect V3 heater from common heater supply and connect to 23/24 winding.
Parallel 25/26 and 29/30 windings (new common heater supply), 25 connected to 29, 26 connected to 30.
Connect 25/29 and 26/30 supply to common heater group.
Re-configure V24 from feeding negative side of negative rail to positive side, cathode to ground.
Instal silicon bridge rectifiers in front of V22, V23 and V24.
Instal voltage doubler in front of V21.
Connect V21 circuit to winding 14/15.
Connect V22, V23 circuit to winding 11/12 in series with winding 18/19, 11 connected to 18.
Connect V24 circuit to winding 16,17.
Connect V24 heater to common heater group.
Instal 5V/4A transformer for 5V4 rectifiers.

Don’t know what all the fuss is about really! The whole scheme works perfectly with the transformer remaining cool to the touch, both the laminations and the windings.
The 316 transformer even fits the chassis cut-out and bolt pattern!
Here is the transplant (I have not yet finished re-lacing the wiring harness), the few components mounted to the power transformer tag-strips are a new calibrator generator circuit described below:

For comparison and the record, here is the original power supply schematic:

Here is a picture showing the Tektronix 316 transformer with the Hammond 5v transformer on top of it. Tektronix and Hammond in the same breath?**! I did my best to sneak the wires down below and drilled 4 holes in the chassis at the gap between the Tek transformer core and the back of the chassis. There was not room to use grommets so I sleeved the wires used thick-walled teflon tube. You may also see the Y amplifier bias shift lever switch I referred to earlier in the narrative at top centre, the knob is missing as is so often the case with lever switches.

Having made so many changes, I now decided to sort out the Y calibration issue. The existing “calibrator” was a switched attenuator followed by a potentiometer for continuous variation fed from the heater supply or, in the case of the high range, a dedicated 34.5v winding on the power transformer. There are two problems with this:
The actual voltage is anyone’s guess since the line is voltage is not stable and these days, the line wave-form is not sinusoidal so calculating the peak-to-peak value is anyone’s guess and who wants to be bothered with that anyway?

Tektronix recognised these deficiencies and soon remedied the issue by introducing an accurately settable clipper to provide a square-wave of known amplitude. To this end, I decided to create a 10VP-P bi-directional clipper using zener diodes. It just happens that the 225V supply includes a 37V winding, all I had to do was to work out a capacitor feed scheme for the zener clipper to DC isolate it. I decided to keep the existing potentiometer on the output of the calibrator however the value of it at 5k meant that I had to create a lower impedance circuit than I really wanted. The result shown below came after much fiddling with values to obtain a good square wave driving the 5k pot. The 2k feed resistor gets slightly warm to the touch. To get good accuracy requires selecting the zener diodes, I got lucky and had to make only one change. The result is a decently accurate 60Hz square-wave at 10VP-P, 1VP-P and 0.1VP-P levels, better than 2%. Each level can be continuously varied to zero if you feel so persuaded; this feature is actually useless, however I was not willing to make any cosmetic changes to the front panel. Even I have some scruples.
Here is the schematic:

Here is the resulting calibrator display, the Y axis at 2.5V/cm. If you are sharp, you will notice that the time setting is slightly off for 60Hz, while the setting at “1” on the dial is pretty good, the dial itself is not accurate, I guess the pots have aged, remembering that the marks on the dial are unique to each pot:

Here is the right side, again showing the transformer adaptation:
Here is the underside, I have re-laced the main harness. If you look at the top right, you can see where I have snuck the 5v heater transformer wires down. You may also spot the 3-section Tektronix timing capacitor bank, it is not original; a previous custodian had fitted it, in the process bending the tags and breaking the seals! I replaced it with another one having short stubby pins that do not foul the casing:
Here is the left side:
Finally, here is the top. Some gibbering idiot took some scotchbrite to the CRT warning label. Unlike Bumblebee capacitors, I did not find it necessary to replace the Black Beauties.
All-in-all, this example is now far from original. Given that the CRT label had already been ruined along with the power transformer, I feel that what I have done with it is worthwhile and most certainly better than the usual alternative.