Early Tektronix (535) & HP (150A) also Tek Type 535A
Here are the Tektronix 535 and the Hewlett Packard 150A both displaying the fast output of a HP 211A square-wave generator, terminated in 75 Ω at a sweep rate of 20 nS/div. The Tek is faster at 36 nS compared with the HP at 44 nS.
Tektronix introduced the concept of top performance main-frame oscilloscopes that support Y axis plug ins designed for specific areas of application. The 530 model series, that was introduced in 1954, was the debut of mainframe scopes of which the 535 model includes dual time bases to permit extremely accurate sweep delay; this feature permits very accurate measurement of the duration of fast events. The Model 535 oscilloscope was last featured in the Tektronix catalog in 1958 at a price of $1300. Think about that!
Starting with the premier league, let me say a bit about how Tektronix came into existence: Howard Vollum worked with the “radar guys” in Britain during the war, alongside the likes of Alan Blumlein; much of the work involved innovative fast, triggered pulse circuits and he came back to the USA with his natural innovative brilliance fully stimulated. He was also awarded the Legion of Merit. At that time, the leader in the oscilloscope market was Allan Dumont and the instruments were not capable of accurate measurement of the events displayed. In particular, the X axis would have a repetitive “sweep” generator, synchronised to the Y signal under observation at the Y signal frequency or a sub-multiple thereof so that 1, 2 or more cycles were visible and stable. This is OK for observation of the form of the signal but in order to make any time measurements, some form of marker from a crystal controlled marker generator would have to be superimposed (usually by beam blanking) so that the time intervals of the signal could be interpolated. This is a tedious method and lacks precision. Howard, with his knowledge of triggered pulse circuits designed a triggered oscilloscope whereby the sweep generator sweeps one sweep at a time when triggered by the Y signal. By arranging for the sweep generator to trigger at exactly the same voltage level each time, a stable display was obtained. The big difference is that the sweep rate or velocity in say, mS/cm, was known, so that the period of the Y signal could be read directly from the graticule. The first commercial instrument was the 511, introduced in early 1947 followed by the 511A in 1950. These instruments (and indeed, instruments they now had become) were expensive and Allan Dumont expressed scepticism, he was wrong, this was a breakthrough and the world wanted it!
Very soon after the 511A, came the 531 and 535 plug-in oscilloscopes; these are arguably the first recognisably modern oscilloscope. My 535 works SO well, is extremely capable and a real pleasure to use. Dick Ropiequet and John Kobbe had developed a much superior wide-band time-base for the wide band 315 project. This time-base was known as the “Ropiequet Sweep Circuit” even though John Kobbe worked with Ropiequet on it; John Kobbe went on to further refine it for the 531/535. This circuit embodied all the functional blocks that we now recognise as essential for a high-performance triggered time-base. These are: A sensitive and fast trigger that responds to a repeatable voltage level on the signal under examination the level of which that can be adjusted using the level control; A gate multi-vibrator that turns the sweep generator on and off, also providing the unblanking gate for the CRT; A linear sweep generator; A disconnect diode that disconnects the sweep generator from the gate as it ramps so that re-triggering cannot happen during the sweep (I have observed this problem with the 511A I have); A sweep length circuit that sends a signal to the gate multi-vibrator to cut the sweep generator off at exactly the same point on each sweep; a hold-off circuit that locks the gate multivibrator out from the trigger until sufficient time has passed for the time-base capacitor to fully discharge. All this results in a stable, reliable, accurate time-base. The 531 had a single timebase; for the 535, this superb time-base was duplicated to create a delaying (B) timebase, used to delay the onset of the primary or delayed (A) time-base. A delay pick-off circuit was created that like the trigger, responds precisely to a voltage level. It can be set to fire at an exactly repeatable level of the delaying sweep ramp, triggering the delayed sweep so that the system can be set up to hugely magnify an event being observed by running the delayed sweep many time faster. The delay pick off can be used to place the start of an event at an exact point on the graticule and the reading on the multi-turn delay vernier noted; then the delay can be increased until the end of the event comes to the same location and the second reading noted. Since the vernier is calibrated accurately in time, subtracting the two numbers gives the duration of the event, that has been effectively magnified several screen diameters. With the 530 series and specifically the 535, I say that we had arrived at the modern oscilloscope, a scientific instrument. Nobody else was even close at this point as you will see when you read about the slightly later HP offering. It did not have time delay, much less calibrated delay and was unreliable to the point of being an acknowledged embarrassment to HP.
To begin, here are some pictures of an early Tektronix 535 serial number 1115, that I found by the side of the road complete with cart and plug ins around 1998. It was this find that planted the seed for this hobby. Tektronix started their production serial numbers with 101, making this the 1014th unit. The manual I have came with unit 8311 and the schematics in it are dated 1957 so I think it is likely that my unit came from the first year of production, 1954 and thus three years older than me. It has the squared-off one-piece case finished in brown, not the “Tektronix” blue that many will be familiar with. I have restored it, which starting with stripping and washing it, including removing the cardboard sleeved electrolytic capacitors, together with all the “Bumblebee” capacitors (which will need to be replaced, more on this later) and the high voltage supply board. I also removed the CRT shield to make access easier. I used undiluted simple green applied with a soft-ish paint brush, rinsed off with 4 gallons of distilled water and dried in the hot sun. Some people say that the use of distilled water is not necessary however, it is cheap and so why take the risk of power transformer failure? Some water WILL get into the power transformer.
Notice the superlative ceramic strip construction. This set a new standard and enabled Tektronix to produce very complex machines that were (are) extremely reliable. It contains 60 valves plus the CRT while the Type C plug in contains a further 12 valves making for a total of 73! I don’t have a Type C, the later Type CA I have contains 15 valves. (The primary difference between the Type C and the Type CA is that the latter has an extremely useful algebraic addition function.)
Bringing it back to life involved testing all the electrolytic capacitors and re-forming most of them (check out the Electrolytic Capacitors post for more information on this) and replacing ALL the “bumblebee” paper in oil capacitors. These capacitors ALWAYS leak, without exception and can cause frustrating problems for the non-initiated. Even if the circuit works, where these capacitors are used for timing, the timing will be slow due to the leakage current robbing the charging current. In the high-voltage supply, the tank capacitor will cause oscillator failure and in the regulated power supplies, they will cause the voltages to be low. Other than that, I spent much time sorting through valves. I do use a tester to screen out exhausted valves (on the Hickok tester, you can use the life test procedure) however, the main thing is to find valves that work properly in the circuits and a tester really doesn’t bring much here. Tektronix advises to go by whether a valve works in the circuit, not by tester results, my experience reflects that which given their brilliance, should come as no surprise. The two areas of performance that arise most often are: transconductance in the trigger valves and power supply regulation error amplifiers and balance and drift in the amplifier valves. For twin triodes in the differential amplifier stages, balance is the key, both current and transconductance together with drift and for individual amplifier valves (most likely pentodes), transconductance matching and drift. DC drift and trigger instability can be due to small amounts of grid current. All this takes patience but if you wish to calibrate your restored instrument, taking time here is well worthwhile. For the common 6BQ7 / 6DJ8 types, I buy lots of several used on eBay and sort through them. There is much rejection however, paying for an expensive, tested single valve will often result in disappointment, again, testing on a tester doesn’t bring much to the game here.
The high voltage transformer turned out to be unusable, fortunately, I had a spare on hand. I understand that this is atypical for the early units which were sealed in beeswax, I guess I just got unlucky. I had removed the high voltage board (with transformer) before washing. I also had to replace all the 5642 high voltage rectifiers. The leads on the ones that were there were mostly broken off at the glass seal, rendering them unusable. The high voltage board was a mess, clearly there had been a traumatic event here! eBay again came to the rescue with several used and unused examples.
In 1956, Hewlett Packard introduced a machine having a plug in Y system, intended to compete with Tektronix (presumably, the 535). I have one of these, serial number 641, which the US vintage HP authority Kenneth Kuhn considers to be possibly the only extant working example other than the one in the HP Memory Project collection of Marc Mislanghe located in south western France. It had been in private ownership for some 50 years which may explain why it remains operable.
When I received it, there was evidence of considerable work done many years ago, before the last owner purchased it. The first task was to replace all the high-voltage capacitors (available from http://www.justradios.com). The power supply overload circuit would keep kicking in, shutting the beast down and and I traced that down to a hugely out-of-tolerance sensing resistor. Now it would work, for a bit. The next thing was sorting out why the beam kept fading. What was happening was that the (seperate) grid and cathode high voltage supplies would drift apart. I note from the manual that HP revised the circuit twice in the first year. Kenneth told me that he thinks the high voltage problem was the main show-stopping fault in the design. There are others… I came up with a scheme to tie the two supplies together without affecting the sweep unblanking. I was also not impressed with the stability of the triggering. Studying the design and comparing it with that of the Tektronix 535, I noted that the Tektronix design picks off from both phases of the Y amplifier thereby reducing susceptibility to common mode artefacts so I modified the trigger pick-off to a design similar to that used in the Tek 535. Purists beware, I used a t**s*st*r for an emitter follower since I needed another active element to replicate the 535 design in this scope. The other problem I experienced was that the time base would not run consistently, I eventually traced that to a leaky mica HF compensation cap between the sweep gate and the sweep integrator.
With perseverance, it is now usable and is actually nicer to use than the Tektronix scopes, for me, at least. The display is very sharp and clear and the controls have a nice, smooth tactile feel that is lacking in the Tektronix units that I have. It contains 49 valves plus the CRT while the 152A plug in contains 10 valves, making for a total of 60.
Having turned it on and off literally dozens of times as I wrestled with the various problems, I had some concern about the life of the CRT. A search for a 5AMP CRT brought up many responses, none of which were for this tube! Browsing eBay one day, I came across what appeared to be a NOS 5AMP2 starting at $9.99. I was the only bidder. The seller graciously followed my packing instructions at no extra cost and for less than $25, a pristine Raytheon 5AMP2 arrived. Here it is:
Below is a comparison of the primary features of these two instruments. In many ways, there is not a lot to choose between them. Apart from Tektronix offering a wider range of plug ins (or the hardware so the user could build a custom one) than HP, the major difference is that the Tektronix mainframe features a second (delaying) B time base that may be used to delay triggering of the A time base by a precise amount of time. This permits extremely accurate measurement of fast events on the signal under examination. The delaying time base is used to start the A time base (set at a suitably fast rate) at the start of the event of interest. This permits the event to be expanded such that the whole event may occupy many screen diameters. Since the time delay is accurately calibrated, the time duration of the event may be measured by noting the delay time at the start of the event, then increasing the delay time until the end of the event is located on the graticule where the beginning was, and then subtracting the two readings; multiplying the difference by the delay time gives the event duration.
Another feature of this system is that in the “delaying sweep mode”, the delayed sweep is arranged to be brighter than the delaying sweep so that rotating the delay time multiplier causes the part of the trace to be expanded to show bright so that the user can adjust the two sweep speeds to “bright up” the event of interest; then switching to the “main sweep delayed” mode displays the brightened portion of the trace magnified at the faster speed. For example if the user brightened the leading edge of an event, he could switch to the “main sweep delayed” mode and measure the rise time. Further, by noting the location of the start of the event on the graticule and rotating the delay time multiplier, he could bring the end of the event to the same graticule location and accurately obtain the event duration.
The HP does provide up to 100X magnification of events but if you want to measure the duration of an event, this magnification is only usable up to the point where the event occupies just one screen diameter. I believe HP thought that they had “got” Tektronix with this feature however, the Tektronix dual time-base architecture actually permits even greater magnification AND you can roll the end of the event that is effectively off the screen by whatever amount of time into view and be able to measure that time, accurately. In other words, the Tektronix approach was/is in a completely higher league to that of the HP sweep magnifier.
|Feature||Tektronix 535 with C Plug In||HP 150A with 152A Plug In|
|Dimensions||13W x 16.75H x 24D||13.75W x 17H x 24.75D|
|Number of Valves||60 (Excluding CRT); C Plug in, 12||49 (Excluding CRT); 152A Plug in, 10|
|Display||5” CRT, 10kv post deflection acceleration||5” CRT, 5kv acceleration (Not PDA)|
|Dual Channel||A, B, invert A/B, Alt, Chop||A, B, invert A/B, Alt, Chop|
|Bandwidth||DC – 10MHz||DC – 10MHz|
|Sensitivity||0.05v/cm to 20v/cm||0.05v/cm to 50v/cm|
|A Sweep Rate||1S/cm to 0.1µS/cm||5S/cm to 0.1µS/cm|
|Magnifier||5X (Max sweep rate 0.02µS/cm)||1X, 5X, 10X, 50X, 100X; (Max sweep rate limited to 0.02µS/cm)|
|Multiplier||Up to 12X (Min sweep rate 12S/cm)||None|
|B (Delaying Sweep) Rate||10mS/cm to 2µS/cm||None|
At a quick glance at the above data, it would seem that there was not much to choose between these instruments, it takes actually using the dual time base system of the 535 to really grasp the difference. Perhaps the testimony of Jim Williams (in his book, Analog Circuit Design) is most telling, he says “The Tektronix 535. Introduced in 1954, this vastly superior instrument made a mockery of all the competition.” He further said “The 535’s engineering concepts and production execution were a bumpless combination of interdisciplinary technology, inspired design, attention to detail, aesthetics and usability. It combined solid knowledge of fundamentals, unbounded thinking and methodical discipline to produce a superior product. The thing just radiated intellectual honesty.” Who am I to agree? Well having stripped, washed and reassembled this one from a neglected and pillaged mess and then seen it simply WORK, restorable to proper calibration without any fuss, I do feel that I am qualified to say yes, I completely agree. This, in contrast to the 150A which was really difficult to restore to proper function. In fact Dave Packard had this to say about the 150A “We decided to develop and build an oscilloscope. We did it on a crash basis; because of the time involved, we did a lousy job.” These two statements, one on the Tektronix 535 by one of the most respected analogue designers ever, and the other by Dave Packard on the HP 150A, really sum the difference.
Below are pictures showing the HP on top of the Tek. The difference in quality is obvious to me at least. You may compare the mechanical construction of the 150A with that of the 535, in particular the use of phenolic printed circuit boards in the 150A versus the use of ceramic strips in the 535.
Note, clicking on each should result in the picture displaying to fit your browser.
Here are some pictures of the 152A and CA Y plug ins, once again the cleanliness of the Tektronix offering is apparent when compared to the HP offering:
In 1959, type 535A was introduced having better controls and a slightly faster vertical amplifier at 15MHz versus 11MHz for the 535. Moreover, the A timebase control was upgraded from a 8 position range switch with a 3 position multiplier switch that provided 24 settings to a single 24 position range switch, the speed range remaining the same at 5S/cm to 100nS/cm. The B timebase range increased from 12 settings 10mS/cm to 2µS/cm to 18 settings, 1S/cm to 2µS/cm. On type 535, no trigger mode switch was provided for the B timebase and in fact, it was necessary to provide an external connection when operating in B delay A mode, you can see where I have made this connection in the picture below. The space vacated by the deletion of the A multiplier switch from type 535 was replaced by a B timebase mode switch in type 535A. These two control modifications were a big improvement allowing far easier manipulation of a superbly conceived and designed timebase architecture.
Here they are shown displaying a 1MHz squarewave from the fast rise output of a type 106 generator, 535 on the left and 535A on the right:
The construction of the 535 was based on a welded frame of angle aluminium and in the 535A, this was replaced by a structure based on 3 fore and aft aluminium extrusions, 2 at the bottom and 1 on the top left side connecting end plates. A forth extrusion was placed on the top centre, fore and aft, for the two carrying handles. The cabinet was also changed from the hard to remove and replace one piece box to clamshell sides and separate base, all of which could be removed separately. The picture below shows the underside of both types and two additional extrusions that were placed laterally to take the weight of the power transformer off the chassis pan in the 535A can be seen:
Here is the top of the two types showing the central extrusion for the handles:
Finally, here is the left hand side showing changes to the Y amplifier:
It can be seen that the original Y amplifier was on two pans, one above the other while the Y amplifier of type 535A is on a single pan. The delay line of type 535A is also more compact.