I have decided to take on writing up at least some of the GR kit in this collection. This unit was for sale at an attractive price given that it included a very neat GR 1201-B unit regulated power supply. It did not work properly, ceasing to pulse whenever I turned the pulse width control much either side of centre. I did not do well in diagnosing the problem, diving in too deep before simply studying the thing. The result was a cycle of picking it up and putting it aside. A very perplexing issue was that the pulse width was off by a factor of ten and I got into all sorts of trouble with that one, sure that it was acting as a frequency divider even though that made no sense. Finally I saw it, a previous muddler had (re)connected the pulse width pot incorrectly, damn! And so it now works quite well. The Pulse Repetition Frequency was low and again, manual in hand went through all sorts of component value conniptions (based on the manual) only to realise upon studying the circuit calmly, that the tube condition would affect the PRF. The manual makes no mention of this yet looking at the circuit it must be, and is. I simply went through a number of tubes until the PRF came in correct at the calibrated position of the PRF pot. Such dependence on the tube is not necessary and is surprising to me considering the reputation GR has.
I don’t have a lot more to say about it. It works well after many the self-made detours. The manual claims that it can deliver rise times of less than 18nS and fall times less than 10nS. I was able to see 18nS and 10nS on my Tek 475 (250MHz bandwidth) but not less. It does find quite a bit of use on my bench.
Repetition Rate, 2.5Hz to 500kHz with calibrated points in 1-3 sequence from 10Hz to 300kHz plus 500kHz. Continuous coverage with uncalibrated control.
Duration, 100nS to 1S in 7 decade ranges.
Pulse Output Levels: + and – 40mA pulses, each 40Vpk into internal 1k load. DC coupled with DC component negative wrt ground.
The 1201-B power supply provides a regulated floating 300VDC up to 70mA and 6.3VAC at 4A. In the pulse generator, the floating voltage is referenced to ground at -150V and +150V.
The pulse repetition generator is a Schmitt trigger with a RC circuit charge/discharge appended. The C of the time constant is charged from the plate circuit of the A section via the R of the time constant that is between the plate and the grid, thus the charge and discharge of the grid drives the A section. If the A section is off (plate high), the capacitor will charge through the R until the A section turns on*, turning the “B” section off. This action happens suddenly due to the regenerative switching action of a Schmitt trigger. The B section has an inductor in its plate circuit and the sudden release of energy due to the B section turning off produces a sharp pulse that is used to trigger the output pulse timing circuit. When the A section turns on, its plate voltage will fall and the capacitor will discharge until the A section turns off again. In this way, the oscillation swings takes place within the hysteresis of the trigger. The reason for PRF dependency on the tube as well as the RC values is that the hysteresis of the circuit will vary from tube to tube. The PRF ranges are controlled by switching in different values of C and continuous control is provided by making R variable from the front panel.
* The manual incorrectly states this as off. Here is the repetition generator circuit:
V101 serves as a current source when the circuit is in generate mode (switch position O), allowing the Schmitt trigger to freely swing around its hysteresis point. In position A, it acts as a driver for the Schmitt trigger to allow the circuit to be driven from an external source.
The positive trigger pulse from the repetition generator is applied to the pulse timing circuit and then the whole bloody thing goes belly up, or at least it did until I spotted the wiring blunder.
Q101 is normally on and with it, V103A and diode V103B, holding the voltage at the junction of R118 & R122 at a level determined by the setting of R125, the Pulse Duration control. Since Q101 is on, V105 is also on producing a current in the positive pulse output load resistor, R130.
A positive pulse trigger pulse from V102 (the pulse repetition generator Schmitt circuit) turns Q101 off and with it, V105 and V103A. Q102 turns on, and with it V106, producing a current in the negative pulse output load, R133. C2 begins to charge (ramp up) via R118 and the grid voltage of V104 rises until this tube conducts and the Schmitt circuit V104A and B changes state, sending a positive triggering pulse to Q102, turning Q102 off and Q101 on, re-establishing the initial circuit conditions. The circuit is now ready for the next cycle initiation pulse from the Pulse Repetition Generator. The pulse duration is therefore controlled by the time constant R118, C2 and the initial potential on C2 established by the Pulse Duration control.
The pulse output circuit consists of pentode V105 and V106 that are switched as described above by the transistor bi-stable, Q101 and Q 102. In the initial state, Q101 is off and with it, V105 causing the output pulse to rise to ground, going low when the circuit changes state. V106 operates in the opposite direction. The screen voltages set the zero bias currents at 40 to 45mA and since the output loads are 1k, the pulse amplitudes are 40 to 45 volts in maximum amplitude.
I have one of these neat power supplies each for my 1217-B Pulse Generator and 1210-C RC Oscillator so I thought I would do a quick post on it. It supplies a floating precision 300V at 70mA and 6.3V at 4A.
A search on the net did not bring up the schematic for this unit so I have traced it out:
(P.S. After doing this I did find a link to the manual for this unit! Go here)
Apart from the primary control loop from the output to the series pass tube via the error comparator (12AT7), amplifier (6AN8 P) and cathode follower (6AN8 T), there are two other signal paths: C531 and R533 feed input fluctuations forward to the grid of the series pass tube in cancelation phase via the amplifier (6AN8 P) and R540 serves to make the open loop gain infinite thereby allowing very low output impedance. The technique of regenerative feedback to allow infinite open loop gain coupled with degenerative feedback to realise extremely solid stability and freedom from drift is also used in the TS-375 A/U VTVM DC amplifier which is the subject of my previous post.
I haven’t characterised the impedance vs frequency however, the use of a grounded grid pentode amplifier with a cathode follower to drive the grid of the series pass tube clearly indicates that it was designed to have excellent performance at high frequencies. My own regulator designs usually include feed-forward (DC coupled) but I have never attempted to make the open loop gain infinite.
Here is the inside with the PCB flipped up:
INTRODUCTION: This unit potentially has practical value to me since it is both an AC and a DC Valve-Voltmeter, also as an AC meter at 100MHz plus, it has a much wider bandwidth than my AC only HP 400H (4MHz). (By the way, Valve Voltmeter because I am of English origin and I think that valves are SO much more interesting than toobs, you see.) I have had this VVM for quite a while and it was one of those projects that I flirted with and then set aside, again and again. The reasons for this were 1, it did not work properly and 2, I could not understand the circuit; I finally re-drew the DC amplifier and meter circuit for clarity and I have included this diagram in the theory section below.
CONDITION: It is in excellent condition with tubes that are effectively free from microphonics and balance perfectly, and the balance once set, stays put. However, I could never get it to read accurately and consistently. I had removed the meter cover and gently blown on the needle, it seemed completely free however, I finally realised that it was in fact sticky. The disturbing force due to blowing is quite a bit stronger than that due to the available magnetomotive force! Since the meter is quite well sealed, I felt that the problem was not likely to be particles in the gap so I gingerly backed off the top pivot. Sure enough, the meter cleared and now does move under electrical stimulus freely and can be reversed at any point without sticking. I cannot offer any explanation as to why the pivots were dragging other than perhaps corrosion. Having said that, I do wonder at the quality of the (Simpson) movement; The mass balance of the needle is very poor despite the presence of balance weights, and it can only be calibrated and used either flat or vertical. I set it up flat because I will use it on a bench, not sitting on a shelf. The picture below shows the large mechanical zero error due to the mass unbalance:
DESCRIPTION: It is an extremely neat unit, and typically for military equipment is enclosed in a grey aluminium case. (By the way, I do have the knobs from the terminals, I simply forgot to replace them before taking the picture.) Both the AC and DC probes are present however, other than the power cord and spares fuses and lamps, none of the other clips and cords are present. Here it is showing the probe compartments open:
The AC input at the terminals is routed via the rectifier probe which is stored on a mounting clip that connects to the probe tip and probe grounding ring. To make accurate measurements above 100MHz, it is necessary to connect the source signal ground directly to the probe ground ring and test tip directly. Even at 40MHz, the leads should not be longer than 3 or 4 inches, according to the manual. The probe tip and ground ring with their respective cradle contacts are both shown in the picture below. The ground wire should not be more than at the most 4″ long. This means that the source ground and test points must be close together. One problem was that the tip of the AC probe was shorting onto the case, you can see where the paint has been scratched. I managed to adjust the cradle a bit but it is still close:
Here is the spares compartment:
And here is what is contains:
And finally, before we get in deeper, here is the inside, note the two spare knobs and spare rectifier on the left by the pots:
AC and DC voltage ranges, 1.2, 3, 12, 30, 120, DC only, 300.
Input loading, DC 30MΩ, that is useful for me!
AC, 5MΩ shunted by 70pF at the panel terminals or 5pF at the AC probe tip. The manual has a graph showing that the resistive component of the loading falls (the load increases) with frequency, two points from the graph being 5MΩ at 10KHz and 90,000KΩ at 100MHz. A good scope with a top quality X10 probe can do better, typically 10M shunted by 2pF. However, when this unit was produced, scopes that could match the bandwidth of this meter existed only in engineer’s dreams! (Engineers are weird like that, I KNOW!) Also, this unit is much more handy than a lab grade scope of the era in any case.
DC, all ranges 3% FSD.
AC, 10 to 50Hz, 5% with correction curve
50Hz to 100MHz, 4% without correction
50MHz to 150MHz, 3% with correction curve
150MHz to 300MHz, 8% with correction curve
The meter meets the % FSD specification on the AC and DC ranges. I made at least 3 spot checks on each range after some tweaking. It is worth noting that as is normal for meters, the specification is given as % FSD; The % of actual values is often quite large but within 5% except for the actual value at 1V on the 3VDC range which is -6.25%, this translates to -2.08% FSD. Most DC errors were on the low side which suggests that I might be able to set the calibration better. Having said that, the AC errors were a mix of high and low so to actually accomplish better calibration would be very tricky and most likely not stable, so I will not attempt to “improve” it.
It is essential to carefully zero the meter with the input terminals shorted on each AC range, you cannot simply switch from range to range. The zero variation on the DC ranges is negligible. This is very likely due to diode contact potential about which I say more in the theory section below.
I checked the frequency response. It is claimed to be flat over the range 100Hz to 100MHz which if true is excellent, for the period of this meter at least. I checked this aspect out using a HP 8601A generator terminated into a 50Ω through BNC with the AC probe and ground ring connected right at the termination, on the 3VAC range I observed a 1dB increase up to 97MHz increasing to +1.5dB at 110 MHz which is the generator limit. Clearly there is a resonance somewhere in the signal path yet I am impressed. This meter is better than it seemed to me at first, second and third blush. This is the connection arrangement to the signal generator:
Overall, this seems to be a capable instrument, at least in the context of its vintage. However, it is essential confirm to that it is reading accurately in the range you desire to use by connecting a known accurate meter in parallel with it and testing it using a DC or true sine AC supply. This is not as silly as it sounds because the point of this meter is the extremely light load it applies to the circuit under test and its wide bandwidth. Your DVM probably cannot match it in these aspects.
THEORY OF OPERATION:
Essentially, it is a DC amplifier and meter. A good DC amplifier is not a trivial proposition because of drift. Most VVMs rely on large amounts of degenerative feedback to stabilize the circuit, however this can result in problems with HF oscillation. Perhaps that is why the HP 400H mentioned above is limited to 4MHz. This design relies on both regenerative and degenerative feedback that I will attempt to explain with the aid of the manual. (The HP 425 Micro-Volt/Ammeter uses a light chopper that produces a rough squarewave that is proportional to the DC current and in this way, completely removes drift. This is followed by an amplifier having a passband at the chop frequency that suppresses other frequencies (including 60Hz) heavily. In fact the chop frequency is chosen not to be harmonically related to line frequency to further ensure no hum on the extremely fragile signals. The amplified chop is then demodulated, rectified and applied to the meter. I think this exemplifies that DC amplifiers are truly non-trivial!)
Because it is a DC device, AC currents must be rectified and this takes place in the AC probe. The result is a peak voltage and this is corrected such that the resulting DC voltage presented to the DC amplifier is equal to the RMS voltage, if the AC current is truly sinusoidal*. This correction is applied by R-111 in TS-375/U and fixed R-133 plus variable R-134 in TS-375A/U. If the waveform is other than sinusoidal, a form or crest factor must be applied. The AC rectifier is right at the probe tip so that the AC signal does not have to travel along a wire and this is how the wide bandwidth is realised.
* The GR 1201-C oscillator that I was using to test and calibrate the AC ranges with has a slight clip on the positive peak (I don’t like GR equipment other than their bridges, heresy I know) and this was sufficient distortion to confound my efforts. The HP DVM I used as a reference would indicate the true RMS of this signal but the VVM would not. I ended up using a HP 205AG that produces a clean waveform with enough available amplitude to test the 120V range.
First here is the simplified circuit:
The screen cross-coupling provides the regenerative paths while the DC plate to grid offsets, (coupling batteries in the diagram) provide the degenerative paths and the output terminals. The plate resistors and the tubes comprise the 4 arms of a bridge and so you can think of the meter as being across the bridge from one plate/plate resistor node to the other plate/plate resistor node, via the coupling batteries. This visualisation may help you to see the bridge.
Degenerative feedback brings stability, any error is returned to the input as a cancellation voltage. However there is a limit to the degree of degenerative feedback possible with a conventional amplifier due to the limits of the single stage amplification. In this design, the additional gain needed for stability is obtained by regeneration, permitting a higher level of degenerative feedback and thus higher voltage stability in this single stage circuit than would otherwise be possible. Quoting from the manual “By this method a gain ratio greater than the actual gain of the amplifier tubes proper can be realized.”
Now to the actual circuit: The “batteries” have been replaced by gas voltage stabiliser tubes, the internal resistance of the tubes being low enough to act effectively as batteries in this circuit. The strike currents for the stabiliser tubes is provided by the keep-alive resistors, R-106 and R-107, each being 100k. These terminate at the negative supply lead. The keep alive currents are too great for each plate circuit so the positive end of the strike tubes are fed via V-103 cathode followers from the + supply lead, the cathode followers being driven by each screen grid. In this way, the practical circuit emulates the simplified circuit. The resistor R-108 is included in the cathode circuit to allow for the voltage dropped across R-106 and R-107.
I found the circuit as drawn difficult to follow so I re-drew it thus:
The regenerative (positive) feedback paths are obvious, R-104 limits the regeneration to a prescribed level. What was not obvious to me was the degenerative (negative) paths and the meter circuit so my re-draw show how the meter is connected to each plate/plate resistor node of the bridge via the screen CFs and the gas tubes. The gas tube keep alive resistors are effectively in parallel (Thevenin) with the plate load resistors. The degenerative feedback paths are shown in blue and red. Taking the blue first, the path is plate to ground then via the input voltage, to the input grid. Since the whole circuit is floating, the ground connection does not affect the operation. The red feedback path includes the meter which has a negligible resistance of 1K. Any grid current is orders of magnitude too small to cause sensible deflections of the meter which is driven by the current due to the unbalancing of the plate voltages.
(At this point a question arose for me: Why isn’t the meter simply connected directly across the plates? The answer is that the resistance value of the meter path would reduce the regenerative feedback too much. As it is, the CFs isolate the R-104 regeneration path from the meter path. It is a clever circuit.)
At the point of optimum adjustment, the screen-grid circuit is critically regenerated and on the point of self-sustaining oscillation if the degenerative paths to the control grids were disconnected. The manual suggests thinking of the circuit as an amplifier having infinite gain due to the critical regeneration which is completely degenerated by external feedback paths (the blue and red paths); thus the stability improvement possible due to degenerative feedback around any amplifier is in this case, carried to the limit.
At this point of critical regeneration, the screen grids do the actual work of unbalancing the bridge to produce an output (proportional to the input) and the control grids simply serve to initiate the unbalancing action. For any steady value of input, the potential of each control grid with respect to its cathode is the same and the potential difference across the grids is zero. The value of zero grid voltage excursion is that it entirely removes the effect of the curved grid voltage, plate current relationship and the amplifier is strictly linear in its input to output relationship. As I said, it is a clever circuit!
I mentioned diode contact potential earlier: A hot cathode diode will develop a negative potential on the plate with no ac input, this potential is termed the “contact potential” of the plate wrt the cathode. The diode in the probe is compensated by adding a similar diode to the red side grid circuit. I suspect that the very slight change of DC resistance across the compensating diode as the range is changed may account for the need to re-zero the balance every time the ac range is changed.
This is the first item of GR equipment that I have written about. I have been resisting (and continue to resist) GR gear that in it’s way, is superb and accurate, but fussy. The mechanical construction tends to be tricky and every item I have is hard to work on. I think that the electronics must have been designed by mathematicians, not engineers. This is not helped by skimpy, confusing manuals that I need a magnifying glass to read. Given GR’s reputation, there is no excuse for such poor manuals that contain difficult to read schematics with obscure descriptions and instructions that contain many errors. It is almost as though the construction and the manuals conspire to protect the secrets within!
The 1568-A usually comes with an XLR input connector at the top left (the third pin was used to provide a microphone energisation potential), this one has an input level control. (I have marked a full clockwise arrow to indicate calibrate next to this control.) I replaced the God-forsaken 1/4″ jack input with a RCA connector. My experience of the reliability of jacks is dismal.
The GR 1568-A is a wave analyser that operates over the range 10Hz to 20KHz. According to the manual it was “inexpensive”, I take that with a grain of salt. The filter is of the constant percentage bandwidth type where the width of the response peak measured between -3dB points is always 1% of the centre frequency resulting in more than 75dB attenuation an octave away from the center frequency. Again, according to the manual, it is particularly suited to analyse closely spaced components common to the vibration spectra produced by various types of machines (presumably including record cutting and transcription machines).
The analyser consists of three main sections, an input amplifier, the variable peak filter and an output amplifier / meter driver. There is also a precision step attenuator that consists of a maximum input control and a meter range control.
The filter consists of a synchronous cascade of two resonant (second order) filters, the first filter having high Q and the second low Q. This filter system is continuously tuneable from 10Hz to 20KHz, the synchronisation being accomplished by two precision wirewound frequency potentiometers that are geared together.
Well, in a word, tricky, this is my first experience of such a machine and I have found it extremely challenging. In principle, the first thing to do is to set it in calibrate mode at the centre of the frequency range of interest. (The manual does not specify whether centre refers to the log centre of the linear centre. For example for a range of of 1KHz to 10KHz, the log centre is 3.16KHz and the linear centre is 5.5KHz. This seems like pedantry I suppose and maybe if I understood the mathematics properly I would know.) The calibrator is interesting: When in calibrate mode, the output of the filter amplifier is fed back to the input. The forward gain is set using the input and meter range step attenuator by placing white dots on each control over each other at 12 o’ clock. The loop gain is controlled by the cal control and is to be adjusted such that the gain of the analyser is equal to the loss in the feedback network such that the system oscillates, the resulting signal being displayed on the meter. The cal control is operated so that the meter needle lies in the calibrate region on the meter face.
The system then may be adjusted to indicate the fundamental frequency of the signal to be analysed at some indicated reference level, either voltage or 0dB. Then sweep the frequency upwards, noting the frequency and amplitude of the harmonics. It is nice to set it in automatic so that the frequency range indexes automatically. The meter is scaled 1-10 and 1-3 also +2dB to -15dB. If the amplitude of the signal falls too low to be resolved accurately on the meter, the meter range may be stepped downwards in 10dB increments or corresponding voltage increments, thereby increasing the sensitivity by a known amount. The overall range is 100µV to 300V full scale. The analysing range of the system is given as 80dB so for example, if the fundamental is 1Vrms, then the smallest measurable harmonic would be 100µVrms in amplitude. I would think that battery operation is necessary to realise the highest sensitivity of the unit.
The intention of the automatic mode is interesting: That is the frequency control may be rotated by a drive from a GR1521 chart recorder so that as the pen moves across the paper, the frequency sweeps upwards, the range automatically indexing with each rotation. Type GR 1521 is fitted with a drive sprocket. The knob on the 1568 is replaced with a corresponding sprocket and placed on top of the 1521. The drive is hooked up with a chain and set to a slight backlash with an idler sprocket. The days before computer based analysis! It would be fun to see this set-up working, preferably demonstrated by a competent operator, not me!
CONDITION ON RECEIPT:
Completely dead. I expected that the two 9.6V Ni-cad batteries would be defunct however, it turned out that the meter coil was open-circuit. The wire was so fine that I could barely see it using a magnifying glass, and I am quite good at finding broken ends of fine windings and repairing them. It was a Jewell taut-band 100µA device. I scoured the panel meter offerings on Ebay and eventually took a chance on a nice looking 100µA taut band meter, it that looked as though I might be able to transplant the movement from this meter into the GR meter case. I got lucky, it worked out beautifully though the needle is a little short. I find it does not hinder reading the meter significantly and the linearity of the scaling matches the GR scale well. The picture of the analyser above shows it with the rebuilt meter. Here is the original movement:
Here is the transplant:
Another issue was a failed zener in the high Q section power supply, easily replaced. Later, one section of the 4 section precision wirewound frequency pot in the high Q section went open. I got lucky here in two respects, 1, it was the top section than is relatively easy to open and 2, all it took to repair it was to re-flow the resistance wire connection at one end. Phew! Here is the open pot:
Here is a view showing showing the high Q and low Q pots that are geared together. Without doubt I got lucky with the location and nature of the pot fault! NOTE, I took GREAT CARE to place index marks on the potentiometer shaft, rotor and case before disassembly!
Here is an overall view. Partially visible on the bottom right is the potentiometer rotor. The high Q board is on the left, the low Q to the right of it. Underneath the filter boards is the frequency multiplier switch that is rotated via bevel gears (visible in the picture behind the rotary switch) either directly from the front panel or by a motor when the unit is operating in automatic mode. This compartment is completely shielded, closure being completed by a plate over the top. To the right of the filter compartment is the PSU board and above that is the meter drive board. The capacitors are for the 10Hz range with the attenuator between. The two frequency potentiometer shafts are visible at the bottom, the printed switch cuts the filter path when the dial is crossing between ranges and the micro switch on the right indexes the frequency range motor when the unit is in automatic mode or when operating manually and crossing over up to the next range. This mode is not available when the unit is powered from the batteries.
As far as I know, this instrument is unique, at least for the valve era. It was originally made for in-house use, then word got out, it was cleaned up for production and remained in the catalogue for 21 years. Tek people referred to it as “Elsea”.
It provides direct readings of low values of L and C and also provides a so called “guard voltage” that can be used to charge any stray capacitances due to nearby conducting materials. A great example would be measuring the inter-electrode capacitances of valves, the guard voltage is connected to all electrodes not to be measured. It can be similarly used to measure capacitances that are in-circuit. I recently used it to check a 0.5 µH deflection plate coupling choke for a Tek type 547 that I re-made after the original broke while I was disconnecting it. (The small Allen-Bradley resistor core cracked.)
The principal of operation is deceptively simple: There is a 140KHz fixed oscillator and a second 140KHz oscillator that drops in frequency when a component is added at the test terminals. The two frequencies are fed to a mixer and the difference is measured using a capacitance charge circuit that is arranged such that the voltage the capacitor charges to increases with frequency (the difference frequency) and this voltage is directly indicated on a meter. If zero reactance is present, the two oscillators dead-beat (a bit like me really) and the meter indicates zero.
The guard voltage referred to above is simply the variable oscillator output buffered by a cathode follower so that the stray surfaces are charged to the same conditions as the surfaces to be measured.
The only problem it had was oxidised pins on the variable oscillator valve that required some physical agitation together with the Deoxit treatment. After that, it worked properly and I was able to calibrate it.
Tektronix produced a switched box of standard Cs, a couple of Rs, an L and a short circuit that they termed S-30 Delta Standards, for this job:
Here is the inside. It is heavily constructed and shielded. Despite that, the connector side is bent and I chose not to attempt to straighten it:
I attached the standards box to the back of the meter using self-adhesive velcro pads, I have an abhorrence of vital accessories to equipment becoming separated. When it was made, the standards would obviously be held in a calibration lab but now that such equipment is obsolete, I make an effort to keep things united…..
(Please click on this picture to view it properly. I think that the meter on the 190 A is better looking than the meter on the 190 B.)
I acquired a clean 190 B sans attenuator which is essential to the function of this design. One of my friends at VintageTek said that they had a unit on the scrap pile that had retained its attenuator so in due course that unit, which turned out to be a 190 A, arrived here. Even though it was dirty and missing the covers, one tube was missing and another tube had a failed heater (which is rare in my experience) I could not resist checking it out. In addition, two of the PSU capacitors were dry and I initially reformed them, then later, replaced them. Having put in the missing 6C4 oscillator tube and replaced the dead 12AU7, being Tektronix of course it worked, and worked properly too. The unit also had a 4 pin Jones socket mounted on the back, presumably so as to use it as a regulated power supply? I could not tell for sure since the socket had been disconnected.
As luck would have it, an attenuator turned up on Ebay for a reasonable price so I now have a working 190 A and B! As always, I applied Deoxit to all tube pins and switch contacts.
The basis of the unit is a Colpitts oscillator with 5 switched ranges covering 350KHz to 50MHz which at the time, was sufficient to support bandwidth testing of oscilloscopes. There is also a fixed 50KHz output. An output attenuator is provided at the end of a lead that has a male UHF connector mounted on it for direct connection to an oscilloscope. The attenuator is coupled to the generator by a special purpose Cannon connector that has a VHF coaxial connector enclosed with 3 other pins that support feedback of a DC signal that is linearly related to the peak-to-peak amplitude of the HF signal at the attenuator. The attenuator has 7 ranges from 0.1 to 10Vp-p, (a constant variation control is provided on the generator). The attenuator contains diodes that sample and rectify the HF output, resulting in a negative DC voltage that is close to the peak to peak amplitude of the HF signal, and that is linearly proportional to the HF signal. This DC signal is returned to the generator via the special Cannon connecter, and is used to control a regulated power supply that feeds the plate of the Colpitts oscillator and by this means, maintains a constant amplitude at the point of application, the attenuator output. The manual states that if the shunt capacitance at the output is less than 50pF, the output amplitude will vary less than +/- 2% from 50KHz to 30MHz and less than +/- 5% from 30MHz to 50MHz. Using my 200MHz Tek 475, both units appear to be very much flatter than specified. The output impedance is 52 Ohms. Here is the special connector:
The DC sample voltage is also used to drive a calibrated meter that indicates the output amplitude in p-p volts, it also shows when the generator is being operated within its controlled amplitude envelope. The frequency is indicated on a vertical drum with a separate scale for each and an vertical illuminated cursor line in a window on the front panel.
In earlier equipment, the sampling diodes were a dual diode 6110 tube, later replaced in the B version by 1N87 silicon diodes. Both my attenuators have the 6110 tube since the B unit was divorced from its original attenuator. If you are looking closely, you may have noticed an empty hole near the top of the chassis; the A version had a 6AL5 double diode in this location. The meter is connected across the cathodes of a dual triode, one grid of which is connected to the sampling diodes, and the other that was connected to the 6AL5 that is connected in the same configuration as the sampling diodes. The intention was to minimise thermal drift of the thermionic diodes being registered by the meter. I suppose that I could replace the 6110 in the attenuator that I have connected to the B unit with 1N87s but it is more likely that I will instal the 6AL5 compensation circuit instead, if I do anything. Interestingly and fortuitously, the B unit retains the heater supply to the attenuator.
Here is a Wireless World advert for this instrument that I stole ages ago from another website. I can’t remember where I got it from so if it was you, please let me know and I will acknowledge you for it:
The ebay seller originally had this up at $150 which was very high. He also said that he had turned it on and some smoke came out that he didn’t think was serious. Hmmm. So I contacted him and let him know not to just turn on old equipment because damage may well result, up to and including power transformer failure*. (This author knows, he has learnt the hard way.) I later determined that the smoke had come from the negative rail smoothing resistor that was running into a dry capacitor that was a near dead short. The result was that he re-listed it at $75 with free shipping. I was the only bidder so another relic arrived on my doorstep. I must say, the seller was very prompt in getting this to me so I left him positive feedback.
*Notably, the three power transformer failures that I have precipitated were all in USM oscilloscopes (I told you that I know, damn it). My thoughts on this are that 1/ the USM scopes tend to be rather tightly enclosed, presumably to reduce RFI emissions and 2/ this, probably combined with much duty, resulted in prolonged high temperatures hence degraded power transformer insulation combined with failed capacitors. Also, my experience prior to looking at USM scopes was almost entirely with Tektronix. In their early years, Tektronix experienced a spate of power transformer failures that they addressed by bringing in Gordon Sloat to set up their own transformer manufacturing facility. Tektronix also include 10Ω carbon fusing resistors in each rectifier circuit to protect the transformer and this excellent practice may have been inspired by the intention to make not only the best performing, but the most reliable oscilloscopes possible at the time, so I was spoilt. Encountering the very interesting USM scopes certainly increased my knowledge and experience base!
The general condition of the case is good with the exception of the CRT hood which is showing some denting and loss of the black anodising. The prop tilt stand is missing also the detector probe is missing from the accessories. As usual with new acquisitions, I opened it up, to find it in fair condition but dirty and oily due to oil migration from many leaky paper-in-oil (PIO) capacitors, I ended up replacing most of them including the HV capacitors. Most of the POI caps were superbly neat molded phenolic types, it is a shame that the sealing of the phenolic around the leadouts did not stand the test of time and temperature. All the power supply electrolytics were dry. I disconnected each one and reformed them one-by-one. Only one unit failed (the one referred to as causing the sellers non serious smoke), it did reform but the ESR was so high that it was useless. (I understand that this problem can be due to corrosion of the internal connection between the aluminium foil tails and the solder tab and/or can.) Many of the tube screen cans are exhibiting “season cracking” a phenomenon whereby deep drawn brass will transition from ductile to brittle at low temperatures so this unit apart from being hot, must have spent some time at freezing temperatures too!
At this point, the real work began for it was not happy! A hard-copy manual (that I prefer) was not available, my friend Volker Klocke has the manual in pdf form on his website at
It has early printed circuit boards. The tube sockets connect to the traces by side contacts that do not overlap or engage with the traces physically, instead relying on a solder “bridge” from each contact to the associated trace, a recipe for intermittence! I spent much time re-working these connections, this job was made more difficult by the connections on the sockets which were corroded and would not tin easily, I ended up using flux and then thoroughly cleaning each area using flux-off and a stiff brush. The other task was to test the tubes. I usually do not do this however it has been my experience that the tubes in the AN/USM scopes are often exhausted (which tends to confirm my thought about prolonged hot service). In this case, most of the tubes were strong with only two exceptions including the HV rectifier. It was apparent that somebody had gone through this example at some point, evidenced by some truly horrible soldering and melted insulation. Weller soldering gun anyone? I also found that the HV circuit had been re-wired incorrectly resulting in the unblanking multivibrator (that rides on the negative end of the HV supply to allow DC coupling to the CRT grid) not working.
Failure of PIO caps shows up in two primary ways, low power supply voltage due to a leaky bypass cap(s) or a circuit is drawing too much current due to a grid associated with a cap being high. In this case the HV voltage was less than 1000V, down from 1500V and one of the supply regulator gas tubes would extinguish when I switched in the marker generator. It is worth noting that failure of coupling caps can (and do) cause power transformer failure as well as failed electrolytic caps in the power supply!
PURPOSE and outline description
This OS-57, USM-38 oscilloscope is serial number 527 and was manufactured by the Trad Electronics Corp of New Jersey. Along with many of the vacuum tube era armed service scopes this model falls into the synchroscope category whereby the timebase can be driven by an internal trigger generator that has an output on the front panel that may in turn, be used to trigger the circuit being investigated in sync with the timebase. Applications would include radar circuits and logic circuits. Somebody tried to tell me that I was wrong in using the term synchroscope, insisting that this term refers to the well known AC power phasing device. Well yes it does, and also to the triggering/triggered oscilloscope. In fact Tektronix, in their book “Using your type 535 or type 545 oscilloscope”, refers to the term synchroscope when describing how the gate pulse from the B timebase may be used to trigger both the A timebase and the circuit under examination.
Here is the trigger and marker generator assembly:
This model uses the classic (and excellent) 3WP1 CRT with 27 tubes plus 2 hivac neons hence the 30 valve claim in the Wireless World advert above.
The timebase is of the triggered multivibrator type that is ac coupled to a push-pull deflection amplifier that is dc coupled to the X plates. The timebase has 5 ranges, 10mS/in, 1mS/in, 100µS/in, 10µS/in and 1µS/in with the X gain set for a 2.5″ sweep length and the sweep speed turned fully CW. The CRT remains cutoff until the timebase sweeps, to prevent burning of the screen by a stationary dot. A sweep expansion feature is provided that allows 9X magnification of any region of the waveform (why 9X I don’t know).
It has an internal trigger generator, rate variable from 40 to 5000 pulses/S, also a marker generator with settings at 100, 10 and 1µS.
The ac coupled Y amplifier has 5 stages including a pentode long-tailed-pair that provides push-pull deflection, ac coupled to the CRT plates. The front end is a switched attenuator that presents a constant 1M / 40pF load to the input, it has 5 ranges, 1, 3, 10, 30, 100 and 300x corresponding to sensitivity ranging from approximately 200mV/in to >50V/in. A 400nS delay line is also included. A variable up to 1Vp-p calibrated signal is provided that may be switched in; this in combination with a variable Y gain control allows on the spot calibration of the Y channel. The bandwidth of the Y amplifier is 10Hz to 6MHz. A 75Ω dummy load is available that may be plugged into the CF probe socket on the front panel to provide a standardised load to the input.
Accessories that are stored in the front panel cover include a cathode follower probe, a 10x attenuator probe, a detector probe and the aforementioned plug-in 75Ω dummy load.
Timebase. The heart of the timebase is a sweep gating multivibrator that may be adjusted from astable (free running) to monostable (triggered) using the stability control that alters the degree of negative bias applied to the multi. When the bias is sufficiently negative, the multi is held in the wait state until triggered, otherwise it will free-run. The multi provides a negative gate (rectangular pulse) to drive a simple capacitor charge sweep circuit. In the wait state, the sweep generator tube is normally conducting, that is discharging the timing capacitor. Upon receipt of the negative rectangular pulse from the gate generator, that is applied to the grid of the sweep tube, the tube cuts off, allowing the timing capacitor to charge or sweep at a rate controlled by the setting (resistance) of the variable sweep speed control. When the gate multi reverts to the wait state, the resulting positive pulse turns the sweep generator tube back on, discharging the timing capacitor (flyback).
The gate multi has a time constant that causes it to pause to allow the sweep to take place before reverting to the flyback and wait state. The gate time constant is switched with the 5 sweep rate ranges and it is also varied in tandem with the variable sweep speed control, holding the gate time constant at approximately 1/10 of the sweep time constant so that the sweep amplitude is limited to about 1/10 of the charging voltage; since the first 10% of an exponential rise/decay is substantively linear this technique results in good (surprisingly so) sweep linearity. It has the further benefit (for a given setting of the X gain) of holding the sweep length constant.
The sync signal is applied to the gating multi via a coupling diode that passes only the negative going signals that are required to trigger the multi while in the wait state. (The multi is anode triggered in case you spot the apparent contradiction between the negative going trigger and increasing negative bias locking the multi into the wait state.) In the sweep state, a positive going signal would be needed to trigger the multi so that as long as the multi is in the sweep state, it is prevented from re-triggering during the sweep by the coupling diode.
The sweep gate is also used to unblank the CRT during the sweep; the positive (inverted) going gate being capacitor coupled to the CRT grid for HF unblanking. Unblanking at low frequencies is assisted by a bistable multi (referred to as the intensity gate shaper) that rides on the negative end of the CRT supply. It is turned on and off by the unblank signal and provides the necessary square topped positive unblank pulse, directly coupled to the CRT grid.
X Amplifier. The output from the sweep generator is buffered by a cathode follower that drives the X gain control. From here, the positive going sweep signal is ac coupled to the long-tailed-pair X deflection amp. The input grid of the X deflection amp is also connected to the X shift control via a diode. The direction of the diode is such that as the sweep moves positive it disconnects, and at the end of the sweep reconnects thereby restoring the grid potential to the shift potential so that the sweep always starts from the same place, reducing jitter on the display.
Sweep Expansion. The X expansion switches in a further gain stage having a gain of approximately 9. In this mode, similarly to the grid clamp above, the start of the sweep is clamped to ground to prevent expansion jitter. The extra gain stage is arranged with a bias control that causes the stage to respond from the start of the sweep and then as the bias is increased the start point moves progressively up the ramp causing the expanded display to move along the waveform under examination. The result is that any 10% of the normal sweep can be expanded and displayed on the screen.
The timebase may be driven by the internal pulse generator or from the signal applied to the Y axis or from an external signal. There is no separate trigger circuit, the sweep gate is triggered directly by an amplified sync signal from the Y axis or from the trigger generator. The way to operate this timebase is to set the sync full ACW then bring the stability down (turning CW) until the timebase free runs, then back up again until the timebase just stops. Then bring the sync back up until the timebase triggers and locks. It may be necessary to repeat this operation if you want a different sweep speed. Triggering from the Y axis or an external signal may be selected from the rising or falling signal.
Synchronisation Amplifier. The sync amplifier consists of two gain stages and the sync selector switch that allows the user to select triggering from from a rising (+) or falling (-) edge from the Y amp or an external signal, or triggering from the internal trigger generator. The gate is triggered by a negative going signal only so the sync amp is arranged to generate a negative going signal from either a negative or positive going signal; + selection causes the sync signal to be inverted and – selection preserves the polarity of the signal.
Y Amplifier and Calibrator. The 1st stage of the Y amplifier is a pentode video amplifier that is followed by the 2nd stage cathode follower followed by a series resistor from the cathode to provide the required 1k source impedance to the delay line that is terminated with a 1k resistor shunt at the input to the 3rd stage that is also a cathode follower. The 4th stage is a second pentode video amplifier which in turn drives a push-pull pentode deflection amplifier that is ac coupled to the Y plates. The delay line drive CF is also used for sync pick-off to the sync amplifier.
In addition to the constant impedance attenuator at the front end, there is a user variable gain control located between the 1st and 2nd stages that allows the deflection to be calibrated using the internal 1Vp-p 60Hz calibration squareish wave. The calibration signal is derived using a + and – diode clipper from the power transformer.
Power Supply. There are 3 DC power supplies, positive, negative and HV (negative). The 250V B+ supply is of the choke input type and this is the first time I have encountered choke input in any equipment! It is vacuum tube rectified. A tap on one side of the B+ winding supplies a half-wave tube rectified C-R-C filtered 200V negative supply, that is used for the timebase including the gate bias and the X amplifier long-tail. The negative 1500V supply is fed from a 1100V winding that is (as is usual) a continuation of one side of the B+ winding, it ends in a winding that feeds the HV rectifier filament. The half-wave rectified HV is smoothed by a simple C-R-C filter. There is a separate heater winding for the CRT and unblanking shaper bistable multi.
Bottom view, you may spot the ceramic wirewound resistor that replaced the burnt negative rail smoothing resistor that was caused by the seller “testing” (ho hum) the unit with the failed electrolytic: