Genrad 1689M RLC Digibridge repair and calibration
The new keyboard on the left (IET part no. 7880-3200-00) to replace the old keyboard on the right.
Ooops! The 3N164 was not mounted correctly (left picture), thus preventing the RAM from going into low power standby mode. This error would not be found during production test, as the 1689M functions as intended when powered on. However, the 1689M would require frequent replacements of the back-up battery.
|
Repairing the 1689M
The 1689M I acquired suffered from bad connections of the IC sockets, a drained battery for the RAM (resulting in a series of error messages), and a bouncy keyboard. In fact, I had to replace the keyboard with an original one to be able to operate the 1689M properly. The bad socket connections were addressed by merely exercizing the ICs slightly out and back into place. With a new lithium battery the 1689M could now store the calibration parameters, but as the back-up power had already been lost, the readings were all over the place with no apparent relation to the component being measured. Each and every calibration step in the service manual was required to put the 1689M back in operation. Another observation was that the RAM voltage was reduced significantly during power down. The current draw from the back-up battery was about 0.2 mA, far more than the PD444 CMOS RAM would draw while in standby mode. It turned out that the P-channel enhancement (= "normally off") MOSFET 3N164 (Q6), used to set the CMOS RAM in low power standby mode, pulled down instead of being open during power down of the 1689M. Much to my surprise, the 3N164 was mounted incorrectly (see the picture), and I only had to insert the 3N164 correctly to obtain the required standby function. After the change the current draw from the battery was reduced to about 12 µA while the 1689M was shut down. In other words, the previous owner of the 1689M must have had to replace the battery way too often. While investigating the above I found that the 1689M is available in two generations: The first generation used LF347 op-amps for buffering and amplification, and had a linear power supply. The later generation uses LT1058 for these amplifiers, and uses a switch-mode power supply plus linear post-regulators. Late models also appear to use a Dallas NVRAM with integrated battery, rather than a static CMOS RAM plus a separate battery. Interestingly, by 1689M had the LF347 replaced with LT1058 (the date code is a clear evidence for that). Calibration was carried out using the home-made resistors below. The calibration routine also includes a step for compensating for the 38.4 MHz crystal oscillator's frequency offset. However, should you want to take the oscillator's drift and offset out of the uncertainty budget, the solution is to synthesize the 38.4 MHz clock from a 10 MHz laboratory reference. |
The set of calibration resistors, plus an open and short, to calibrate the 1689M.
|
Calibration resistors for the 1689M
The Genrad 1689M service manual calls for a set of calibrated resistors of specific values to carry out a calibration: 24.9 Ω, 374 Ω, 5.97 kΩ and 95.3 kΩ. IET labs sells the 1689-9604 Digibridge Calibration Kit with these values, including a short and an open. One reason to make your own set of calibration resistors is that the kit requires the 1689-9600 Remote Test Fixture for use with the 1689M. The Quadtech 7000-09 Calibration Kit includes resistors of the same values as the 1689-9604 kit, and even features four BNC terminals, which makes it easier to use with the Genrad 1689M. However, you will eventually need to have your calibration resistors calibrated, and DIY resistors using high-quality resistors are then as good as those that are commercially available. Precision resistors with low temperature coefficient are typically not stocked in the values listed in the 1689M service manual, but values sufficiently close can be made, combining resistors such as the bulk metal foil models from Vishay that are typically on stock. The actual value of each calibration resistor is entered into the 1689M during calibration, so there's no need to target the exact values specified by Genrad. The calibration resistors I have were made of Vishay S102C (± 2.0 ppm/°C) and Vishay HZ (± 0.2 ppm/°C), both with 0.01 % tolerance, as follows: 24.9 Ω: 50 Ω // 50 Ω = 25 Ω 374 Ω: 750 Ω // 750 Ω = 375 Ω 5.97 kΩ: 5 kΩ + 1 kΩ = 6 kΩ 95.3 kΩ: 100 kΩ I had all of the calibration resistors calibrated to establish traceability and to be able to make an uncertainty budget. The resistors were calibrated at DC and the Q values required for the 1689M were calculated from modeling the parasitics from the through response using a network analyzer. |