Hughes Memoscope 104D
This is a post that I started many months ago. Now that I am winding up my restoration activities it is time to post it, complete or not:
I probably paid too much for this, but it did come with the manual and these machines appear to be rare and most definitely have a place in history, for it seems likely that Hughes beat Tektronix to the market with a storage oscilloscope. The seller did not know anything about it, sadly it had been pillaged of its 8 12AU7s and 2 12AX7s. (I am giving the seller the benefit of the doubt here.) Fortunately, I had enough strong 12AU7s and 12AX7s on hand to replace them. This example has serial number 635.
It is quite nicely made but not even close to Tektronix standards. The bandwidth is a very limited 500kHz* and the timebase triggering is fussy at best. The timebase does not have a trigger shaper (usually a Schmitt trigger followed by a differentiation CR), this probably accounts for the general fussiness of the timebase. The Tektronix 564 storage scope came out soon after and offered 10MHz, a proper timebase, dual timebases if desired, not to mention a wide range of X and Y plug-ins. It was also usefully smaller. Speaking of the Hughes manual, it is awful, there are several circuit and description contradictions making it difficult to follow the (extremely limited) calibration directions. At the age of 56, even with diopter 2 reading glasses, I need a magnifying glass to read the schematics.
* The WB/4 Y plug in limits this further to 250KHz.
The ESSENTIAL thing with these CRTs is to NEVER place them face down. I don’t yet know on 1/30/2014, if this one is OK.
Here, I provide a description of the principal of operation of the Memotron tube, that I have stolen from the manual and re-written somewhat to hopefully make things either more clear or less muddy.
The Memotron tube has two electron guns, a writing gun and a flood gun.
There is a dielectric storage mesh placed before the viewing screen. The flood gun sprays the dielectric storage mesh with a (supposedly) uniform barrage of low-velocity electrons leaving the dielectric surface at the flood gun cathode potential. The high-velocity electron beam from the writing gun charges regions of the storage surface positive by secondary emission*, thus creating areas which are partially transparent to the flood electrons. The flood electrons which pass through are accelerated to high velocity producing a continuously visible image of the positive electrical charge pattern stored on the dielectric surface: as long as the flood gun is emitting, the positive and negative potentials of the charged pattern on the storage surface are maintained thus sustaining the image.
A second mesh called the collector mesh interposed between the writing gun and the storage surface is used to discharge the charge pattern on the storage surface the image by briefly lowering the voltage of the collector mesh.
This technology is sometimes referred to as bistable because a given area on the storage mesh will either be at flood gun cathode or collector mesh potential.
*Bombarding electrons can impart enough energy to surface electrons to cause them to break free, electrons breaking free in this way is known as secondary emission.
* 2/2/2014, well I know now that the CRT does store but I suspect it is not in its prime! The contrast is poor and the background brightness has to be quite high before the trace can be seen, more on this in the design notes below. I knocked up a “pinger” consisting of a 3H choke in parallel with a 1µF cap in series with a 10k resistor. I placed about 20Vdc across the circuit to store energy then quickly disconnected it resulting in the following damped oscillation (timebase is 20mS/div):
Contrast Enhancement Oscillator.
There is a blocking oscillator that applies +15V, 10µS positive pulses at a rate of 1000PPS to the storage mesh, supposedly to enhance contrast by decreasing the background brightness. Though the pulse amplitude is adjustable, adjusting it appears to make no difference at all. This is most likely due to the condition of the tube, however in truth, I have not made the effort to fully understand the action of the pulses on the CRT storage system.
This simply reduces the CRT collector potential to nearly zero at the same time as briefly stopping the contrast oscillator.
This is a simple DC coupled long-tailed pair with gain such that 9v pk-pk on the input terminals will produce full scale deflection of the CRT beam. The amplifier is served by a plug-in that in this case is a DC coupled differential pre-amplifier type WB/4. A dual channel chop/alt plug in WB/D1/11 was also available.
The sensitivity of the WB/4 is 10mV/div to 50V/div with a bandwidth of 250kHz. The legend on the front says “wideband”. Even during the era of this scope, 250KHz was not wideband! The chassis is mounted on rubber anti-vibration mounts however, the sensitivity of the unit does not seem to warrant such a detail. Perhaps this is more important when capturing the slow events that this scope is clearly intended for. I would think that it was used to observe mechanical systems, especially since an aircraft company (Hughes) put for the effort to create it. There must have been a critical need for it. It would be interesting to learn just what is was used for, maybe something, in the arrogance of hindsight, that we now regard as obvious.
This DC coupled amplifier is similar to the Y amplifier except the degree of coupling between the cathodes can be varied to control the gain. The gain is set to cause the sweep to occupy exactly 10 divisions across the screen, best done using a marker generator. The amplifier may accept an external sweep and in this mode the gain may be continously adjusted by 10:1, a further 10:1 being available on the Ext Sweep portion of the Time/Div switch. The sensitivity is 0.5 to 50V/div with a bandwidth of 250KHz.
The timebase is switched on and off (trace, flyback) using a bistable multi-vibrator which is pretty typical practise. However, there is no Schmitt trigger between the vertical signal and the control multi. There is a cathode coupled trigger amp thats allows triggering on positive or negative going signals. The DC output level of the trigger amp is variable so that the trigger signal may be shifted relative to the DC trigger point of the control multi, in this way the DC level on the signal at which the timebase starts may be set of varied. Because there is no fast squaring (Schmitt Action) of the triggering signal followed by differentiation to produce a sharp trigger pulse, this system is unsuited to fast events.
The timebase essentially consists of a bistable control multi and a positive going sweep generator. The multi is anode triggered with the triggering signal being applied via a diode (so that it is negative going) to the input tube that is off, awaiting a trigger. This forces the anode voltage down, this change is communicated to the grid of the on tube causing it to cease conducting and the multi to flip states and the off (input) tube turns on, the anode voltage falling further. The multi remains in this state for the duration of the sweep. The resulting negative pulse is used to turn the timing capacitor discharge tube off and the capacitor starts to charge. Linearity of the charge process is ensured by bootstrapping the charge to a control tube across the charge resistor, maintaining the voltage across the charge resistor constant and thus the charge current also. The sweep voltage rises approximately 90V and is communicated to the grid of the off tube of the multi, causing it to turn back on resulting in a positive going pulse at the grid of the sweep discharge tube, turning it on, and discharging the timing capacitor. With this action, the input tube of the multi is now returned to the off state and the circuit is ready to receive the next trigger. Delay between the discharge process and the next sweep to ensure that the discharge is complete is accomplished by an RC time constant on the grid of the discharge tube causing it to turn on a little after the multi flips back to the sweep state. This is a simplified hold-off, in better designs the hold-off time constant is switched to different values with the sweep range switch. There are 6 ranges, 10µS/div to 1S/div together with a 1X, 2X and 5X times multiplier such that the range of the timebase is 10µS/div to 10S/div. There is also a 10X continuous multiplier control.
This is a clever solution to the problem of coupling the unblanking signal (positive pulse from the timebase control multi) to the beam control (usually grid) element of the CRT. What Hughes did was to turn a self excited oscillator on and off using the unblank pulse. The oscillator runs at around 10.5MHz. The oscillation is fed via a low impedance transmission line to a receiving coil and capacitor that is tuned to the same frequency as the oscillator. The resulting voltage is doubler rectified and applied to the grid of the CRT. Neat.
This consists of a HV oscillator driving an air cored transformer at around 300KHz. A sample is taken from the rectified output and via an amplifier is used to control the screen voltage of the parallel oscillator tubes thereby regulating the output in the usual way. The PDA voltage is 4.3KV while the grid and cathode supply is around -2.85KV.
“Low” Voltage Supplies.
Per Tektronix (and everybody else after Tek showed the way), there is a precision -200V rail, this is used as the reference for stacked +200V and +450V rails. There is also an unregulated +325V rail.
The first job was sorting out replacements for the missing tubes and using Deoxit on all the tube sockets. Then came replacing burnt out power supply protection resistors, a couple of the caps required re-forming and most probably replacement soon. Once this was done, the regulated supplies came in a little low but all in tolerance, the drop-out point was well below line voltage. It was not long before I realised that something was not quite right with the HV (CRT) supply which is a 300KHz high frequency oscillator type with the oscillator amplitude being modulated by amplified DC error feedback from the output to the oscillator screen grid regulator. It uses 2 1X2B rectifiers and on checking them, the negative supply rectifier was very weak so I replaced it too. I was then able to get the negative (-2.85kV) and PDA (+4.3kV) supplies in tolerance though the regulation is not solid and the voltages do vary a little as the intensity control is operated. I can find no obvious reason for this, all components are good and the screen regulator amplifier tube is good too. Here are a couple of views of the HV PSU:
I particularly like the air-cored, pie-wound HV transformer and the intensity and focus control universal joints. The rectifier heaters are supplied by the loops around the base of the transformer.
Interestingly, while the PSU reservoir caps were OK (well, after reforming 2 of them anyway), I did find it necessary to replace the regulator output shunt caps. (Caps that are not designed to handle rectification charging currents are less robust, insulation-wise.) One of them was tucked at the lower back corner of the PCB with the power transformer on the foil side so I had to remove the power transformer from its mounting. If you click on this picture, you will just be able to make out the offending cap behind the wiring in the lower right-hand corner:
And here is the transformer side with the plug-in cage, the CRT and shield and the HV PSU removed:
If you click on the picture, you will be able to see that the power transformer is off its mountings and actually could be swung right out of the corner to reveal the foil pads for the offending capacitor. While you are there, look for the nice universal joints near the front panel that drive insulating rods that connect to second universal joints on the intensity and focus pots in the HV PSU.