AN/USM Oscilloscopes

This AN/USM-8E/U is one of only three repetitive sweep scopes* (i.e. not-triggered) in this collection (the others are the Triumph 830 wobbulator and the Knight KG-630), it was manufactured by Carol Electronics Corp. of Martinsville, WV under navy department – bureau of ships contract number, NObsr-75143. I believe Hickok (of tube tester fame) also manufactured a similar oscilloscope under this contract. It uses a 3RP1 semi-flat faced tube. The 3RP1 is a neat tube but suffers from low sensitivity.

There are links to other USM related posts at the foot of this post but I would like for you to read this one first!

(*Those of you who have not experienced older scopes, may not be familiar with repetitive sweep. This type of sweep existed prior to the introduction of triggered time bases and persisted in cheaper scopes. A repetitive sweep system simply consists of a sawtooth generator synchronised to the signal being displayed. Being synchronised, it will only lock onto signals having a frequency an integer multiple of the sweep frequency, to display any frequency means fiddling with the sweep frequency (repetition rate). As such, a repetitive sweep is not a true time base since it cannot be calibrated in sweep velocity (e.g. 1 mS/cm) and so time measurements cannot be made directly. A well-designed triggered time base will stay locked as the displayed waveform is varied in frequency resulting in the number of cycles being displayed rising rising and falling with frequency and if the sweep velocity is calibrated, time measurements of the waveform can be made directly from the graticule. If the number of cycles is too high or too low, you simply switch the time base to a higher or lower sweep rate. Dick Ropietquet introduced the first true calibrated time base for the Tek 315 model in the early 50s. Since then, typical triggered time bases, mostly modeled on the Ropietquet design, have a sweep rate accuracy of better than 3%.)

Click here to see these two oscilloscopes in action demonstrating the fundamental difference between repetitive and triggered sweep.

This unit arrived in mostly excellent condition. I had to disassemble the Y gain pot because it was stuck. The problem turned out to be a disintegrated phenolic coupling inside which I was able to fabricate a replacement for.

The construction is very neat. I particularly like the way each sub-unit is designed to come out. Top left is the X amp, Top right is the Y amp, bottom right is the synchronising circuit and bottom left is the sweep generator.

Having repaired the Y gain pot and done the usual cleaning of pots, switches and tube pins (using Craig products), it then worked, after a fashion. I then made some modifications (see pictures below for details) to the Y amp and sync take-off. I took care to leave original parts in place so that it can be returned to the original configuration if a future custodian so desires. I also placed a copy of the notes I made on my modifications inside the unit and in the manual.

Here is a picture showing the unit valiantly displaying a 2.5MHz sine wave. Notice the good sweep linearity.

The second AN/USM scope is a OS-34/ USM-32, developed and made by Dumont. It was also sold to the civilian market as the Type 301-A. It is a very compact unit being 7.5 H x 5.5 W x 15 long, excluding the front cover which contains many accessories including a AM detector probe and  triode connected 5702 cathode follower probe. The CRT is a 3WP1 flat faced type that was produced by Dumont amongst others and is notably used in the excellent Tektronix 310 and 310A. It is around twice as sensitive as the 3RP1 used in the OS-8E/U. This scope was intended for specialist use in maintaining radar equipment and in some ways is the most interesting scope in this collection. Note; if you are considering one of these, know that there is no auto or free run capability so when using it as a regular oscilloscope, a trace will only be present when a signal is present (no base line) which is a bit disconcerting.

The design is unusual, for me at least. It features a fairly crude triggered time base that is not calibrated, despite the use of an active shunt-regulated power supply. The shunt regulator is also unusual, all the other scope regulated power supplies I have encountered are series-regulated with the exception of the DC heater supply in the HP 150A. (Technically, shunt-regulation is superior, however, it is vulnerable since if the load is removed, usually the shunt element over-dissipates.) The voltage reference is also a bit quirky, the 5651 reference tube being located in the ground end of the CRT divider chain providing -87V. This means that the divider chain is designed to run at around 2mA which is much higher than necessary than is necessary to properly bias the CRT. Here is my hand re-drawn time base schematic, I did this to help me understand it while getting it working. I found C507 to be leaky, a very similar fault to the one I found in the HP 150A time base. The waveforms and voltages shown were taken after the modification described further down this page.

A crude marker generator is included that blanks the beam at the marker intervals, however it is not crystal controlled so I do do expect any useful accuracy. (As received, it is actually disconnected, I expect the coupling cap is leaky and that was the simplest way to get the scope usable at the time…) The other rather interesting feature is the provision of a trigger generator that produces trigger signals at intervals from around 25mS to 180μS. I say around because that is the specification, there are repetition rate marks around the control but the actual repetition rate did not even come close. I fiddled with it until I got the end points in the ball-park however, the increments are still wildly off. I wonder if ageing of the pot track has changed the rate of the pot as well as the over-all value? The trigger generator puts this unit in the category sometimes referred to as a “synchroscope”. The idea is that the trigger generator is not only used to trigger the time base, but also the external circuit under examination. The intended purpose of this scope was for use in trouble-shooting triggered circuits that might be found in radar and since the unit is naval, I guess also sonar. I am looking forward to getting the unit operating fully and playing with this feature!

This scope also uses the clever unblanking bistable method that I explained in the Solartron CT 436 discussion. In summary, the unblanking signal (from the sweep gate) is ac coupled to a bistable that is riding on the CRT cathode and switches the CRT grid on and off in response to the leading and trailing edges of the sweep gate thereby obtaining proper unblanking at low sweep speeds without DC coupling to gate to the CRT grid.

Here it is as received from via eBay showing the glue used to hold the green filter that has run down over the filter over the years. The only other picture of an OS-34 / USM-32 I have found (at www.navy-radio.com/test.htm) shows the same problem. I simply removed the green filter, it could easily be replaced with a new one for authenticity.

Since this scope was designed for specialist naval requirements to support radar operation, I suppose that its role was critical. In the manual the use of oversize fuses in the event of fuses blowing in an emergency is described (but no recommended). If you look closely you will see a lamp socket marked “Heater” right under the screen. I thought this was for standby operation however, it actually is for a pair of heaters inside to unit to keep moisture at bay! It was intended that the heaters be on continuously except when the unit was in use. The “heaters” are actually a pair of wirewound power resistors across the line feed.

I soon discovered that the power transformer had failed. I had it re-wound by Gary of TRS (xfrmrs@roadrunner.com). At that point, having removed the transformer, I washed the thing using simple green and hot water. I now have it on the bench and am checking out the various circuits using a bench power supply. I determined that the likely cause of power transformer failure was a dried out electrolytic that I gutted and re-stuffed. I also found two open RF chokes! One was “normal”, the failure being where the fragile wire wraps around the pin. The other was quite different and frankly looked like a deliberate cut on the outside of one of the pie windings. I was able to tease out the ends and re-join them.

Here are pictures of the end cap showing the accessories:

The first picture shows layer one (like a box of chocolates) containing the cathode follower probe. The second picture shows layer two containing the demodulator probe and BNC to banana adaptors.

The re-wound power transformer courtesy of Gary. The primary is updated for 120V.

Left side view with trigger, time-base and marker sub-chassis swung out to reveal the re-wound transformer installation:

Left side, PSU caps (including re-stuffed electrolytic), Y system, 110V shunt regulator and trigger generator:

Top:

Back, reference tube, rectifiers, Y deflection tubes and un-blanking bi-stable (vertical tube):

Working! (Somewhat.) Highly staged set-up in “Synchroscope” mode using the internal trigger generator to trigger the time-base and a Tek 106 square-wave generator. You can see the “Trig Out” connection at the bottom of the picture. Note the 5-pin cathode-follower probe socket directly under the “Vert Sig” socket.

There is much to do to get it working reliably and frankly, I am not sure the time-base triggering will ever work truly solidly. I suspect that it will always need finessing to obtain a stable display. Having said all that, it was severely dead when I received it!

I managed to get the time marker working again. Unlike the trigger generator, the rate or period markings do bear some resemblance to reality, the 10mS markers shown here coinciding with 95kHz, pretty decent. The marker works by blanking the beam at a specified period (here 10mS) so if the period of the displayed wave is the same, the beam will blank the same portion of each cycle.

Here we can see that the period of the displayed wave is different to to that of the marker, each cycle is blanked over a different portion.

I think that is pretty much it. The scope is not very stable. The trace moves around with line variations quite badly, mostly I think due to very fickle triggering causing the time-base to start at a slightly different place on each sweep. This, I think accounts for the poor focus. I don’t see myself doing anything to improve triggering accuracy, I think that will require replacing many components for more stable parts. The scope is now demonstrable as and such, a good exhibition piece to demonstrate the technology of the era; around the time Tektronix was getting going, this scope would have represented the state of the art. The quantum leap that Tek made is all too clear.

Modifications: After all, I have spent some time fiddling with the trigger circuit. What I found was that the gate multi-vibrator (which is a schmitt trigger) was bouncing, somewhat like relay contact bounce. There were no grid stoppers anywhere in the sync amp, or trigger/time-base so I went though and added these everywhere using Radio Shack carbon comps (82Ω). This improved matters a bit. I then replaced the trigger cathode resistor with a pot and fiddled with that until the triggering was clean. The value I ended up with was larger thereby increasing the hysteresis or dead-band resulting in stable triggering. The problem now was that the signal from the Y-amp was a bit weak, not quite enough to reliably overcome the increased hysteresis. I noticed that the point of take off had been chosen to minimise the effect of the sync wiring and circuit on the Y-amp bandwidth, however it also limited the amplitude. What I did was to change the location to a point having more amplitude however, less capacitance drive capability. I got around this (as I have done with other ancient scopes) by committing the heresy of inserting a simple BJT emitter follower. This can not only can drive the sync circuit input capacitance but also, by providing a low source impedance, means that the wire from the sync take-off to the sync amp is much less sensitive to hum and noise. This has resulted in a major improvement in usability of the scope. It is now possible to vary the input frequency on any sweep rate setting and see the number of cycles gather and fall from left to right, as it should. As I suspected, the increased trace stability has resulted in an improvement in focus. Oh, I also found that the fine sweep rate pot measured around 30M instead of the intended 5M, no wonder I was having trouble with it! I have since replaced it.

I was still not satisfied though; the time base was over-triggered on large signals causing two issues: One, that the sweep length would contract with larger signals or increasing the “sync” amplitude and two, large signals would be impressed onto the sweep as positive going bumps or ripples (*see note on time base waveforms below) causing the positive peaks of the displayed waveform to be displaced to the right with respect to the negative going peaks, i.e. the wave form would be squewed. I overcame this by placing limiter diodes at the time base driver grid (V501A). Triggering the time base requires that a negative going pulse is applied to the grid of the gate multivibrator tube V504, driving it together with the sweep generator  tube V505A to cut-off which permits the timing capacitor to commence charging. To obtain this negative going pulse requires that a positive going pulse be applied to the grid of V501A. I inserted a 10k series resistor before the grid of V501A followed by a germanium diode, anode to ground to clip off the negative going portion of the signal almost completely. To retain good HF performance, I paralleled the 10k resistor with another germanium diode to conduct the positive going portion of the signal into the Miller capacitance of V501A and clipped that off to around 1.5V peak using three silicon diodes in series, cathodes towards ground. Now, the trace length no longer varies with trace amplitude.

Below are the waveforms with the timebase driven by a 10kHz signal displayed on a Tektronix 561A with a 3A74 Y plug in. *Notice that the trigger is superposed on the sweep gate however, the gate is “deep” enough with respect to the (now) limited trigger that the trigger does not cause the sweep generator to conduct and therefore, the trigger pulses do not appear on the sweep.

Bottom to top: Trigger, 1 v/cm; Sweep Gate, 20 v/cm; Sweep, 50 v/cm; Unblanking, 20 v/cm

I broke down and gave in to my impulse to provide global power supply regulation. Some 40V is dropped across the choke, plenty of headroom for a mosfet. I created a zener 305V reference and placed the mosfet across the choke with the gate connected to the reference. I played with the current distribution between the choke and the mosfet by placing a resistor in series with the choke, ending up at about 60 mosfet, 40 choke. Now, the B+ varies about 1.2% when I vary the line from 115 to 130V. Unregulated the variation is around 35V.  Ripple and noise is less than 5 mV p-p. This modification has further increased the triggering stability and general usability of this curious and obscure oscilloscope.

Details of the various modifications I made to my unit are shown below:

Another picture of it in action showing a transformer ringing. I have installed a new green filter so the scope now looks the way it was when new. Note the solid state HP 180 based AN/USM 281 next to it!

Other AN/USM posts include:

AN/USM-24 Synchroscope
AN/USM-38 synchroscope
TS-452 HF Mechanical Wobbulator
HP (Solid State) AN/USM-281