[Phil G4SPZ]

I bought this neat-looking instrument on impulse as I passed the Bring and Buy stall on my way out of the NVCF in 2007. I paid the princely sum of £3, and so I wasn’t too disappointed when I got it home and found it wasn’t working. Forum member Bill (WME_Bill) kindly replied to my request for circuit information and not only sent me a copy of the user’s manual and circuit, but also his own notes and circuit sketches which subsequently proved to be very useful. A quick look inside revealed a broken solder joint on one of the meter amplifier transistors, and a dab of solder got it working again. However the calibration wasn’t good, but the wide range and sensitivity on AC made it handy for basic tasks such as alignment, where only a peak or a relative reading was required.

Fast forward nine years to last week, and I recently needed a sensitive and accurate AC voltmeter to set up a tape recorder. I started using the Linstead M2B, but it soon became apparent that not only were some of the AC ranges miles out, but the darned thing was highly frequency-dependent. I wasted some time chasing a non-existent equalisation fault in the tape recorder before I began to suspect the meter. This is the second time I’ve been led up the garden path by faulty test gear, so I thought “never again” and started on this repair.

The Linstead M2B was also marketed by RS Components as their “Wideband Millivoltmeter stock no. 610-512” and my example was clearly from that source. The meter came with an accompanying sheet of hand-written notes and calibration measurements, together with a message dated 2007 from a previous owner which read: “I made a case for this meter, it was left over from a student project in my college”. His measurements showed that the calibration was indeed miles out, varying between 16.7% low to 5.3% high on the DC ranges and anything up to 12% high on the AC ranges. It’s obvious why the meter ended up on the Bring & Buy stall the same year!

I started to chase the cause of the frequency dependency. On AC, the input attenuator operates in three stages and includes two pre-set trimmer capacitors, so as I had no setting-up instructions I found by experimentation - using an audio oscillator source - that adjusting these trimmers markedly improved the frequency linearity. I also experimented with the setting of VR2, the ‘set gain’ control in the AC amplifier, and quickly managed to obtain excellent accuracy on all ranges. I don’t have access to lab grade instruments, but I based my calibrations on a Marconi TF1370A wide-range R-C oscillator and I managed to get the Linstead to agree to within a couple of percent. The makers claim an accuracy of +/- 2.5% from 10Hz to 500kHz. Patting myself on the back, I started to put the meter back in its case whilst it was switched on, and noticed the needle swinging wildly as I moved it. In fact, almost any movement or flexing of the chassis caused the meter reading to change. I found and re-soldered two dry joints but the problem persisted. The design is to some extent flawed in that several connections to chassis rely on mechanical tightness of nuts and bolts associated with the wafer switches, which are not of the most robust type. In the end I added three additional wires to electrically bond the crucial points solidly to chassis, and the fault was cured.

Having fixed the AC side so quickly, I rashly assumed that the DC side would be just as simple. Famous last words... on test, the lowest 120mV range was spot-on, but all the other DC ranges read low between 1.5% and 17%. As the instrument was accurate on AC and the two-transistor meter amplifier was common to both AC and DC, I studied the DC attenuator circuit in more detail, and here Bill’s notes came in very handy. The resistive attenuator is again in two stages, an input multiplier and an output attenuator each having four steps, and they operate together in a rather elegant manner to cover the eight DC ranges. I started checking the values of the associated resistors. The input multiplier resistors were within half a percent, but those fitted in the output (meter movement) attenuator differed markedly in value from the circuit diagram.

I worked out that the basic accuracy of the instrument was set on the lowest 120mV range by the input multiplier, as the output attenuator is shorted out entirely. On higher ranges, the output attenuator works by switching various resistors (R34, R37 and R38) in series with the 50uA movement’s negative lead: zero resistance on 120mV; 5.4k on 400mV; 22.8k on 1.2, 12 and 120 volts; and 76.8k on 4, 40 and 400 volts. The input attenuator is a simple resistive multiplier of total resistance 11.23 Megohms, operating in four steps of x1, x0.99, x0.099 and x0.0099. Between the two attenuators the eight DC voltage ranges are covered. The 120mV range was perfectly accurate, and the input multiplier resistor values were all correct, so I tried experimenting with various resistances shunted across those fitted in the output attenuator. Working upwards through the ranges, I temporarily linked a one-megohm pot across R34 and adjusted the pot to give accurate FSD on the 400mV range, then measured the value of the pot at that setting and soldered in the nearest preferred value fixed resistor. I did the same in turn with R37 and R38. By the end of the process I had all ranges within less than 2% of FSD, and I have noted on the circuit diagram the resultant resistances that I used in the output attenuator. Success!

With so much space available, the layout of most of the components is crammed around the two range and function switches S1 and S2, making repair tricky. Also the switch wafers are rather delicate and flimsy - I think these are in the style of ‘Maka-Switch’ components - but using the long tie rods behind S1 to double up as the sole mountings for the AC amplifier PCB leads to a very flimsy construction. The photos show the components strung around the switch assemblies as well as the grounding wires that I added. The resistors that formed the original DC output attenuator and that I had to change are shown at the end of S2.

I can’t deny I was happy to put the instrument back together! It’s amusing to speculate about this meter’s history, as it was obviously hand-built from a kit by a student, albeit with a fair degree of skill. Did RS Components market these instruments in kit form, I wonder? I feel sorry for the student who built it, because the component values supplied for R34, R37 and R38 would have produced gross inaccuracy on the DC ranges. My copy of the circuit diagram also omits a crucial wire link between R34 and the 400mV switch position, which makes understanding the operation of the circuit doubly difficult. Perhaps the student thought that a constructional error had caused the fault, and failing to find the reasons for the poor accuracy, eventually gave up? Anyway, it’s a very useful instrument and one that I can now rely on for accuracy.

Phil