And five weeks from that above post of mine, my version of this d.c. voltage calibrator is now complete.
In my version, I have retained the basic concept of two voltage sources that balance each other: one of those sources supplying the current to the instrument under test / re-calibration. However, most of the rest of the cct. is different.
In the original design, the 2k5 ohm resistors in the resistor chains are under-rated. They dissipate at least 1-watt each and as such their resistance value will tend to change as they get hot. Those changes are enough to require frequent and persistent re-setting of the two 250 v.d.c. rails. Plus, over a period of time, those changes may become permanent: not what you want in a voltage reference. In my design, those resistors have been replaced by 4k7 ohm resistors and the other Rs. in each chain have been changed accordingly.
For the reference 250 v. supply, I didn't use the VR150/30 voltage stabilizer approach as in the original design, nor the C-L-C filter after the rectifier. I used a simple C-R-C filter, with larger capacitors and a linear regulator cct., based on an ECL86, to provide that reference d.c. That enables that reference voltage to be set precisely at the front panel and also monitored. For the balance 250 v., I used a simple Zener diode shunt regulator, again with its 250 v. rail adjustable and monitored on the front panel. The original centre-zero micro-ammeter of the original design has been retained to provide an additional check on the equality of the two voltages: reference 250 v. and the balance 250 v.
A comment on the linear regulator cct., based on an ECL86. In an application like this, although the voltages, current and power dissipation of that valve in this application are within its limits, I have never cared for the idea of supplying large +ve voltages to such valves before their heaters are up to temperature. So, in this design, the H.V. a.c from the mains transformer is fed to a delay circuit at power-on and thus to the two half-wave rectifiers, etc. That delay is adjustable: currently set as 45 seconds. The delay cct. consists of a simple mains-to-12v. d.c. PSU and a purchased timer module, (which was remarkably cheap for what it is
), based on the ubiquitous 555 timer.
Heat control: I regard that as an important consideration in an item of test equipment such as this. The case for this unit is divided into two regions: the rear half contains most of the things that get really hot, particularly the mains transformer and the ECL86. That case half has two fans fitted: one for inlet, one for exhaust. The reduction in temp. of the valve and transformer with those fans is substantial, compared to otherwise.
Usage: the calibrator's use is simple. Initially, a 45-second delay for the timer, then a wait period of about 15 minutes is advisable for things to settle down and stabilize. An adjustment of the two 250-v. sources is then made, checking each on the switchable, suppressed-zero voltmeter on the front panel and adjusting those two voltage rails, again at the front panel, for exactly 250-v. each. A final check on the balance meter is also made.
Construction: I'm not a skilled carpenter! And a close exam. of the wooden case shows that - but that case is adequate for the job and all the wood was salvaged from off-cuts and bits of scrap left over from other jobs. 95% of all the parts were available 'in-house', which was not only convenient but also helped keep the cost down. The top lid of the case and the front panel are each hinged to permit easy maintenance (which, hopefully, will be very infrequent!
)
Results: On test, all the selected voltages at the O/P terminals were found to be within 0.5% of the correct values, as verified on my two calibrated Fluke voltmeters. (N.B.: I'm not aiming at NPL standards here! Only what will be 'adequate' for my own personal use.)
And finally, it was used 'in anger': an AVO 8 Mk. 5 was giving substantial erroneous voltage indications. It was fixed by replacing several high-value Rs. and then checking the resultant indications using this calibrator.
Photos of the finished item are attached: they're not brilliant, but then I'm not a professional photographer, either!
Al.