Time Electronics 2004 voltage calibrator

[EC8010]

It has just arrived and appears to work. It was made in April 1988. It is in better cosmetic condition than I expected; the case isn't dinged or scratched, although there's plenty of missing paint on the front panel's black bezel, which is easily removable, and so might get repainted. If it's very lucky.

Front panel is secured to the case by four M3 screws. Obviously, I opened it up before applying power. One screw was 25mm long. Obviously, someone lost an original screw when opening it up and was incapable of concealing their error.

There are essentially two parts to the instrument, a 14-15V supply is buried inside a folded steel box on the rear panel, and everything else is on the front panel. The two are connected by red and black wires having PP9-type battery connectors on the rear panel's PCB. Opening up the supply reveals tie-wraps that secured a battery about the size of a packet of twenty cigarettes. There is also a two-pin connector that presumably went to the front panel battery monitoring meter. There is a hand-written note on the inside of the back panel:

Set charge
Switch over voltage to 19.100 volts
Charge current = 23mA (crossed out) 45mA
Standby current = 6mA

The battery meter on the front panel shows zero. Mains comes in on a standard IEC connector that appears to be a later modification. The modifier thought insulating tape was appropriate on 5VA mains transformer's bare mains terminals. The same person appears to have replaced the two front panel toggle switches. Insulating tape was wrapped around them to conceal the damage they caused to wire insulation.

Despite the Time Electronics check sheet suggesting that the supply produces 14-15V, I saw 23V unloaded and 17.3V loaded. The power supply PCB has a 7818, producing 18.15V, so I suspect that's part of the battery management, and the main PCB operates from the raw supply.

The wafer switches are RS "Make-a-switch." There are Mann Components 0.01% wirewound resistors on the most significant digit's switch. There are four Spinal Tap wafer switches (they go to 11) plus a variable resistor that doesn't have an end stop (presumably, it had one once), so it is a bit better than five digit resolution, and goes lower in voltage than my Fluke 341A. Each switch is a single wafer, so it's not a Kelvin-Varley divider, and probably adjusts the feedback resistor of an inverting op-amp. There are zillions of trimmers; each of the three most significant digit switches has a trimmer per fixed resistor. I presume that one calibrates by starting at 1 and sequentially adjusting trimmers as one works through switch positions.The output has four ranges of 10V, 1V, 0.1V, 0.01V, presumably achieved by changing the input resistor to the inverting op-amp. A meter monitors output current and is showing 1 (on a scale of 1 to 10) when driving 10V into the 10M load of a 34410A. The manual says its good for 30mA, just like the 341A. There are two large potted lumps bolted to the main PCB, I suspect the larger one contains the voltage reference and op-amp, and the smaller one the meter driver.

Setting it to 10V, I see 10.00010V on a 34410A; I'm not going to be rushing to tweak internal trimmers.

Front panel indications suggest that the instrument can be used to compare voltages up to 300V. Presumably, by disconnecting the links on the front panel terminals. I'm not fiddling with that until I know how to operate it properly...

Next, I'll get the camera out...

[tggzzz]

So, sounds better than I feared, but not as good as you hoped.

Time does tend to pot their voltage references.

What type of trimmer? I've seen horrible exposed skeleton trimmers in Time things.

I suspect the "lots of trimmers" goes with the "grade" spec. Not something I would want to twiddle.

Switches going to "X" (e.g. ESI DekaPot/DekaStat) is good; the reason is obvious once you try to use them! Is it worth cleaning the RS switches before doing much more?

I'm surprised about the high output impedance. The only other time I've seen that is on my Fluke 515A, 3kohms at up to 100V, on exposed banana sockets. What's the current indicated on the 34410's Hi-Z input?

[EC8010]

Agreed, not quite as good as I had hoped, but fortunately, it works. Which rather negates criticisms.

No skeletal trimmers. Mostly the long screw-driven multi-turn type, but the least significant digit to have trimmers has ordinary 270 degree encapsulated types.

Output current meter shows "1" even with external load disconnected. But it shows "0" when power is removed, so there's something internal drawing current after the meter, and there's lots of crashing and banging at switch-on. Poor design. I have not tested output resistance, but I imagine it's low. My low noise bench HT regulator very carefully has the meter after my internal bits and pieces so that it correctly reads zero at zero load current.

Battery was probably ten AA NiCd. The power supply PCB has a 741 and series pass transistor as well as the 18V regulator.

Powering the calibrator from a bench supply set to 17V, it draws 41mA. If I drop the voltage to 14V, current is unchanged and the 10V output is unchanged; there's a regulator somewhere on the large PCB.

The camera's battery is charging...

Meanwhile, I telephoned Time Electronics and very helpful chap kindly emailed me a pdf of the manual and of the specifications. 10/10 for Time Electronics!

[tggzzz]

Agreed, not quite as good as I had hoped, but fortunately, it works. Which rather negates criticisms.

The definition of "works" isn't simple, of course.

I think I would check the linearity by comparing the calibrator output with that from a decent KVD.

As for absolute voltage, well that definition varies occasionally

No skeletal trimmers. Mostly the long screw-driven multi-turn type, but the least significant digit to have trimmers has ordinary 270 degree encapsulated types.

That's fine, and much better than I've seen.

My primary portable voltage reference (a vintage Transorb II) has a 10 turn pot for adjusting the output voltage. Naturally I haven't touched the sticker over it.

Output current meter shows "1" even with external load disconnected. But it shows "0" when power is removed, so there's something internal drawing current after the meter, and there's lots of crashing and banging at switch-on. Poor design. I have not tested output resistance, but I imagine it's low. My low noise bench HT regulator very carefully has the meter after my internal bits and pieces so that it correctly reads zero at zero load current.

Weird. Is the current meter meant to be used as a null meter, e.g. when comparing 300V?

Meanwhile, I telephoned Time Electronics and very helpful chap kindly emailed me a pdf of the manual and of the specifications. 10/10 for Time Electronics!

Oh, very good!

I wonder if they would object if it was uploaded to ko4bb/bama?

[AVGresponding]

The power supply sounds very similar to my 2003N, which had horribly crusty fat coin-cell NiCd stacks in it.

It wouldn't surprise me if there were some commonalities between the two in other areas too (the 2003N is a 5-d "dc voltage potentiometer" and null meter). Yours probably has a hard-wired mains lead like mine, originally, too.

IMG_20261023_162713063_HDR.jpg

IMG_20261023_162724568_HDR.jpg

[EC8010]

Now for some pictures...

According to the manual, the fifth digit (variable resistor) doesn't have an end stop.

[EC8010]

I have asked if Time Electronics would object to upload, and they very kindly agreed. But this site doesn't appear to support posting PDFs!

I don't have a KVD, but I can compare it to my Fluke 341A or I can measure it on my Hioki DM7275 7 1/2 digit DMM. The Hioki is warming up.

First indications are that it has a +11uV offset on its 10V range. But the Hioki has only been on a short while. I'll leave them both on overnight and see what I think. It's been an excuse to use one of my swanky new measurement cables; tightly twisted PTFE-insulated pair in earthed double braid screen with brass 4mm plugs in foam insulation to mitigate against thermocouple effects. Even swankier, I've used one of my four-wire to two-wire cables and used the four-wire sensing ability of the 2004. Whooo!

The back of your 2003N looks very similar to the back of the 2004. Functionality looks very similar except that the 2004 has low thermal EMF binding posts and four-wire capability. Otherwise looks quite similar. I'll bet the innards are very similar.

[EC8010]

I've had a look at the 10V range. Looks like a gain and offset tweak would bring it most of the way in. I hesitate to fiddle with those individual trimmers to get it inside 0.002%, although you could reasonably argue that the standard deviation would already be inside 0.002%.

[EC8010]

I left the 2004 and the Hioki DM7275 on overnight. When I repeated the charted test, it was even worse, with a 352ppm error at 11V. Looks like it hadn't been calibrated in a long time. I followed the Time Electronics calibration procedure, which brought the errors down significantly, but not quite within the original specification of 0.002% (20ppm). So I thought long and hard. Then I tweaked the trimmers on the MSB, pulling the error to a maximum of +/-2ppm on the 10V range going through voltages in integer steps. OK, so this presumes the Hioki is correct, and that its ADC is linear, but it's certainly better than the 2004 as found and should be a little better than an in-specification 2004.

How to do it: The 2004 is effectively an inverting amplifier with its 10V, 1V range switch changing the input resistor, and the five rotary controls changing the feedback resistor. To calibrate each digit, the others need to be set to 0, then start from zero and sequentially tweak the trimmers as you go up. Looking from the back, the trimmer for 1 is bottom right, then count upwards to 5, then count downwards from 6 on the left-hand bank of trimmers. When I checked my first attempt, it all looked good until I reached 6, which had moved, necessarily disturbing the digits above it. As always, backlash in the trimmer means you must wind the same way, then stop. If you overshoot, you have to keep going, then wind it gently back. The Time manual says you shouldn't thump the 2004 and the tweaks I made were pretty small. It was probably thumped quite a bit by Royal Mail, so it's really not surprising it was a long way out of calibration. I normally twitch upon seeing lots of trimmers, but they enabled long-term drift to be corrected.

Having proven the principle, I will have a look at the next two digits later.

Just because I tweaked errors to +/-2ppm doesn't mean it's going to stay there; it will drift. But I've taken out errors accumulated over 38 years and it's now comfortably within its original specification. I like it.

[tggzzz]

The Time manual says you shouldn't thump the 2004 and the tweaks I made were pretty small.

I would have thought there was a tendency for trimpots to "relax" or "settle" a little after being tweaked. Gentle tapping might encourage that process.

[EC8010]

Agreed. I didn't go tapping them, but I did give them time to relax before moving on to the next step, and having worked my way tweaking through the decade, started all over again at zero, but just checking. I have also done the next switch down, and half of the third switch. Annoyingly, there's a big potted lump obscuring the tweaks on one side of the third switch. Fortunately, most of the errors were in the low values on that switch, and as you go down the switches, you can tolerate more error in terms of ppm. I have now finished tweaking and the instrument is comfortably within specification. In all cases, the resistors had gone a little high. I am very pleased with it.

[AVGresponding]

That reminds me, do you know the manufacturer of those green 0.01% resistors? Some searching reveals they are probably Mann Components, a UK manufacturer acquired by Vishay in the 80s. I have some 4-wire ones, and haven't had any luck in getting datasheets

[EC8010]

I do. It is Mann Components of Wymondham, Norfolk, UK (borged by Vishay in 1983) and I have a data sheet (LT series only, I'm afraid) that someone at the other place kindly sent me. I'd post it here, but it's a PDF and this site doesn't like them. I could email it to you if you PM me your email address.

[tggzzz]

I wonder if the traditional manual sites would be prepared to host a copy? Ko4bb, bama spring to mind.

[EC8010]

In my first post, I stated that the back and front panels were secured by M3 screws, but that's wrong; they were 6BA. I have never before encountered Pozidriv dome head on a BA screw, hence my unchecked assumption. I didn't have any suitable (unchewed) replacements, so I ran an M3 tap through the hank bushes and the panels are now secured by stainless steel M3 hex socket cap head screws. The 2004 was designed in 1977, so that was the tail end of the BA era for electronics. There are probably other BA fasteners within the 2004...

[EC8010]

I have posted the Time Electronics PDFs at EEVblog.

[AVGresponding]

I do. It is Mann Components of Wymondham, Norfolk, UK (borged by Vishay in 1983) and I have a data sheet (LT series only, I'm afraid) that someone at the other place kindly sent me. I'd post it here, but it's a PDF and this site doesn't like them. I could email it to you if you PM me your email address.

It's these that I need info on. Have quite a few (gave one to Cerebus in part exchange for some earbuds) of various tolerances and values, but all 4-wire

20230116_172555.jpg