Unit Key Data

[20200517] Measurements

  • The generated (cw) signal was measured with spectrum analyzers [hp70k-02 and hp8663b-01]:
    • Severe frequency deviations (though ‘in the neighborhood’);
    • Inexplicable amplitude reading;
    • Poor spectrum: Very broad (unstable?) signal.
  • This device could use some serious attention (apologies for the glare).
  • Pictures are here.

[20210221] Checking Power Supplies on the Unit

Today, I opened the unit (for the first time, so pictures are available here and here) in order to figure out why the hp8116a output signal is degraded. As usual, after visual inspection for local past explosions or leakage, I checked the power supplies. Most were not too far off, except for the -15V supply; the readout was erratic. I routinely switch my DMM to AC Voltage to have a quick check any ripple on the supply. I’m OK with readings up to a few milliVolts, but the -15V supply was close to a Volt:

I hooked up a scope to confirm my suspicion (excessive ripple on the -15V supply):

Excessive (100 Hz) ripple on the -15V supply of the HP-8116A; vertical scale is 1V/DIV.

I’m not sure if this fully explains the weird, unstable output signal, but having a flaky -15V supply certainly does not help… I called it a day in order to have a look at the schematics and come up with suspects for further investigation. Pictures from the measurement session are here.

[20210222] Thoughts on the Problem with the -15V Rail

Yesterday evening I took the time to study in some more detail the circuit diagram to see the generation and the use of the -15V rail.

The main smoothing capacitor is A1C6, so this is our first suspect. It’s documented value is 470uF. However, with excessive ripple there is always another suspect: excessive current drawn by the circuits on the particular (-15V) rail. And sometimes these two problems are present simultaneously: the main smoothing capacitor has been severely damaged due to excessive load (for instance, due to another leaky filter capacitor). In that case, replacing the smoothing capacitor will not solve the underlying issue.

So, even before de-soldering and examining A1C6, I want to know the current drawn from the -15V rail, and relate that, for instance, to the value of A1C6. Therefore I examined the circuit in some more detail: The voltage across A1C6 is regulated by A1U6 with output-voltage divider resistors (including A1R25 for precise -15V adjustment), further filtered through A1C7 (47uF), A1W5 (zero-ohm resistor ?), A1C113 (47uF) and then sent to the A2 board (only?) through A1W7.

Apart from perhaps through the mysterious A1W5, there seems to be no easy way (like a jumper) to measure the current on the -15V rail. Other options are through A1W7 or through means (if available) on board A2 itself.

Food for thought. But before signing off, I measured (all pictures here) the signal across A1C6 with my tek2440 in the full circuit:

And here’s the same result with A2J4 disconnected, which, to out best knowledge, disables the load on the -15V rail:

There is not much difference with A2J4 disconnected, which supports the assessment that the “-15V problem” already occurs before A2J4. After reconnecting A2J4, we measure:

We see indeed the effect of the load on the -15V on the voltage across A1C6, but the signal only gets marginally worse. It is already unacceptably poor without the A2 load.

This means we can (for now) narrow down our list of suspects to A1C6 (470uF/35V), A1C7 (47uF/50V), and A1C113 (47uF/20V). It seems that luck is on our side and we do not have to check the load circuit…

Time to call it a day…

[20210223] A1C6 is Toast

Today I removed A1C6. I then applied through a diode a negative voltage on the input of the regulator (i.e., on the A1C6 negative pin on the PCB) from a power supply:

I needed quite a negative voltage (~ -26V) to make the ripple acceptable (< 1 mV) on the output of the regulator. At that input voltage, the current drawn from the power supply was 63 mA on a DMM, so no excessive current is drawn. The stability of the output signal vastly improved on the HAMEG scope:

And indeed, through measurement of C6 with a DMM and an ESR Tester, found that A1C6 is faulty. It has close to zero capacitance. However, there are no visual indications of leakage or other problems:

Pictures are available here.

[20210224] Mouser Day (Again)

Today I ordered a replacement for A1C6 from Mouser, along with some other useful electrolytic capacitors. For A1C6 I ordered (10) Nichicon caps with 105 degrees Celsius temperature and 63V (instead of 35V) voltage rating.

[20210305] Replacing A1C6

Needless to say I went immediately into the lab to replace A1C6 once the elcos from the Mouser order arrived. Though the unit was still open, I had stowed it away in a safe place. Before soldering the new A1C6, I took some time to reaffirm the problem with the -15V power supply. Well, it was no dream, the issue was still there.

Pictures of the session are here.

The new A1C6 is 470μF/63V with 105°C temperature rating. Here’s a picture of the replacement job on the component side of A1 (it’s the bottom electrolytic capacitor):

The new A1R6 (the bottom electrolytic capacitor).

And here’s the soldering job:

The soldering job on the A1C6 replacement. The two soldering pads are vertically aligned just left and above the center of the picture. Note that from this picture it actually becomes clear that a few other component replacements have been performed in the past.

After replacing A1C6, the -15V as well as the output signal had improved drastically.

Finally, I readjusted all supplies to be within spec. The series of pictures speak for themselves: I first adjusted the DC voltage, then switched to AC on the HP-3478A for an indication of the ripple. (I ignored the specified A3 voltage for now, as I was not sure were to locate that signal. If it is where I think it is, it is a tad low…) Note that one of the pictures shows on which side of the 0Ω resistors (?) I measured.

Since I was quite happy with the result, I reassembled the unit and did not bother to do a full verification and readjustment. I did, however, have a look at the frequency spectrum of the sine-wave output (1 kHz, 500 mV amplitude) with the hp-70k-02 with DC Coupling (which allows reading a spectrum from 100 Hz upwards!):

The frequency spectrum of a 1 kHz sine wave (amplitude setting is 500 mV).

Note that the second and third (and higher) harmonic are < -45 dBc. Not sure if that’s within spec, but it certainly is enough for now.

I took some goodbye pictures of the original A1C6 and then ditched it.

Unit status -> [operational].

[20210307] Testing Performance

Today I assessed the performance of the HP-8116A unit, following the instruction in the Operating, Programming and Servicing Manual. Before starting, the HP-8116A and all necessary test equipment warmed up for over an hour; the HP-5316A was checked against the (locked) GPSD OCXO, and the HP-3478A was checked against my 10V DC Reference.

Pictures of the session are here.

The results are below: The unit does not meet its specification for a full 100%; in particular not with respect to DC Offset. Moreover, AM/FM Modulation Tests still need to be performed, but these tests require an additional Function Generator. Still, the unit is useful in [operational] status, just minding restriction in the specs on DC Offset and AM/FM Modulation.

01 Frequency Performance Test

I followed the test to the letter using the HP-5316A, and found the following results:

  • 50 MHz -> 49.6 MHz
  • 10 MHz -> 9.79 MHz
  • 10 kHz -> 9.98 kHz
  • 1 kHz -> 999 Hz
  • 1 Hz -> 996 ms [switched input HP-5316A to DC and set appropriate GATE TIME]
  • 100 mHz -> 9.97 s

[Results are within spec.]

02 Duty Cycle Performance Test

For the Duty Cycle Performance Test I used my Tektronix 2440, since I could not find an easy way to measure duty cycle on my HP-5316A. For the 1 Hz case, I captured the waveform on the Tektronix 2440, and used its CURSOR functions to find the duty cycle. For 1 kHz and 9.99 Mhz, I simply used to cursor functions on the continuously-updating display on the Tektronix 2440.

  • 1 Hz Results
    • 10% -> 10.0%
    • 50% -> 50.2%
    • 90% -> 89.9%
  • 1 kHz Results
    • 10% -> 10.3%
    • 50% -> 50.1%
    • 90% -> 89.6%
  • 9.99 MHz Results
    • 20% -> 19.2%
    • 50% -> 50.4%
    • 80% -> 80.7%

[Results are within spec.]

03 Pulse Width Performance Test

With HP-5316A [TI AVG A->B; SEP; A up; B down]:

  • 1 MHz; 100 ns -> 101 ns
  • 100 kHz; 1 μs -> 1.02 μs
  • 1kHz; 100 μs -> 102 μs
  • 10 Hz; 1 ms -> 1.02 ms
  • 1 HZ; 500 ms -> 497 ms

With Tektronix 2440:

  • 10 MHz; 8 ns -> 8 ns. Readout is 7.88 ns, but procedure not that accurate:

[Results are within spec.]

04 Amplitude & Offset Performance Test

This test was performed with the HP-3478A. The DC Offset readings were not 100% stable…

Amplitude (Sine/Triangle/Square):

  • 8V: 2.828 / 2.321 / 4.006
  • 3V: 1.054 / 0.865 / 1.492
  • 1V: 0.354 / 0.289 / 0.499
  • 100mV: 35.0 mV / 28.7 mV / 49.3 mV


  • 100mV: 7.402 [out of spec] / 4.95 / 2.964 / 0.984 / 98.1 mV
  • 1mV: 789 mV / 496 mV / 98.45 mV

[Amplitude results are within spec; Offset results need minor adjustments: just a bit low and out of spec for the 100mV/7.5V case.]

05 Sine Waveform Performance Test

This test was performed by directly connecting the output of the HP-8116A to the HP-8556B Spectrum Analyzer, but only after first assuring that the applied signal would not damage the analyzer input.

Here’s the frequency spectrum measured with the 1 kHz signal:

We find harmonics > -60 dBc at 2 kHz: – 43 dBc (A1), at 3 kHz: -57 dBc (A2); at 4 kHz: -53 dBc (A3). So let’s calculate the Total Harmonic Distortion using the formula given in the description of the manual. We find:

  • 10^{A1/10} = 5.01187233627e-05
  • 10^{A2/10} = 1.995262315e-06
  • 10^{A3/10} = 5.01187233627e-06
  • which sums to 5.7125858014e-05
  • taking the square root: 0.00755816498986
  • multiplying by 100 yields THD = 0.76 %, which is within spec!

And here’s the frequency spectrum measured with the 50 MHz signal:

All harmonics are < -23 dBc and thus within spec; the worst performing is the harmonic at 100 Mhz; this is at approximately -28 dBc.

So, Sine Wave Performance is within spec.

06 Pulse/Squarewave Performance Test

I performed this test through direct connection to the Tektronix 2440‘s first channel with 50Ω input impedance setting.

Here’s the rising edge:

And here’s the falling edge:

The fall time [90->10%] is close to (the required) 7 ns, but from the looks of it, just a tad more…

[The rise time is well within spec; the fall time is almost within spec.]

07 DC Output Performance Test

This test was performed by directly connecting the HP-8116A to the input terminals of the HP-3478A DMM with a 50Ω feed-through. Here are the results; the ones in red do not meet the specification:

  • 7.95 V: 7.854 V
  • 5 V: 4.945 V
  • 2 V: 1.974 V
  • 0 V: -1.86 mV
  • -2 V: -1.984 V
  • -5 V: -4.964 V
  • -7.95 V: -7.861 V

[Some values a bit out of spec.]

08 Trigger, Gate and External Width Verification Test


09 Burst Modes Verification Test (Opt 001)


10 Frequency Modulation Verification Test


11 Amplitude Modulation Verification Test


12 Pulse Width Modulation Verification Test


13 Sweep Modes Verification Test (Opt 001)


14 Autovernier and Output Mode Verification Test


15 HP-IB Verification Test

Skipped, but earlier work on the jinstrument software has already revealed the the GPIB interface works properly.

[20210307] Removing Front-Panel Stickers

While working on performance testing, I removed three stickers from the front panel using my nails, a soft detergent, several screw drivers (used very carefully), and isopropyl alcohol. For the moment, I’m quite pleased with the result, although there is some discoloration:

The HP-8116A after removal of the front-panel stickers and a cleaning job.

More pictures are here.