Test Equipment Anonymous (TEA) group therapy thread

[bd139]

If you go back to the early to mid 1980s when these were designed and look at what was available for backlights you’d see why there isn’t one. The options were EL panels with a lifespan of a few months and limited LED classes leading to wishy washy side lit displays (remember Casio watch backlights). However the desire to have more complex display functions meant that LED displays were unsuitable for production and were pretty expensive both current and device wise to drive. Ergo we ended where we are with it.

Honestly looking at my TF930, in suitable lighting that has a near perfect display.

[mnementh]

Yeah, but a couple 12V grain-of-wheat bulbs from Radio Shank cost a quarter retail, and draw about 40mA.

Every cheap linear power supply with side-view meters had them; no good reason not to include them on several-thousand-dollar test gear aside from not wanting to have to replace them a couple times a decade. Put two in series and feed them ~14V and they last 6-10 years burning pretty much constantly. You could go all-out and spend 25c ea on vibration-resistant automotive dash backlight bulbs with filament supports that can last 20 years.

Yeah, I don't believe there really was a good reason; even the LCD heating issue is easily resolved with a little light-piping design strategy. And we've had illumination-intensity LEDs for 30 years; just required 2 or 4 of them as opposed to a single one like we can use now.

Naahhh; I think it was more a matter of "Just couldn't be arsed to do it."


mnem
I am MOOP.

[Specmaster]

Thinking about it, you're right I do remember those Casio watches with the LED at the side  Compare that with the LCD screens today which have back lighting, chalk and moldy old cheese.

Postman has just delivered my 3478A bits so hopefully that will get done this weekend.

[bd139]

First thing I had with a decent backlight ever was 1995 and featured in this advert

[mnementh]

Yep, that was an early IndiGlo (EL backlit) model; about ten years later they were bought out by the Chinese factory that made those watches for them. That was about the time I abandoned the ironMan line for Fossil; too much corner-cutting, no TIMEX quality.


mnem
The only constant is change.

[Specmaster]

Bugger it, soldering the back battery into place, thought I'd just check that the Cal data was safe before disconnecting the fitted battery, screen says "uncalibrated".

Then it hit me, the solder was touching the bloody chassis as I was making the positive connection to the ram, so now I too have to get the 3478A calibrated, I fucked it 

Oh well I'll continue with the battery and filter caps replacement, then look for a calibration centre near me 

[bd139]

Bugger! Not having much luck this week are we

[tautech]

Bugger! Not having much luck this week are we

We have a saying here for you guys......Penis fingers.......f**k everything ya touch !   

[bd139]

Hahahaha will have to remember that one for the office 

[Specmaster]

Bugger! Not having much luck this week are we

Another reason not to like anything greater than 4.5 digits, like the 3466A it can be user calibrated with ease once you have a reference for each range 

[bd139]

To be fair you can do that with the 3478A. You just need the reference

[Specmaster]

To be fair you can do that with the 3478A. You just need the reference

I might have a go, I have Voltage references and I have photos that show the readings on some 1% reference resistors I have.

Just discovered that my power transformer was loose and only had 1 screw holding it, fixed that now, reassembling after chucking out those fiddly filter caps, especially the fucker next to the transformer, that was a pig to replace due the transformer and the power switch, its a real snug fit down there. 

[Specmaster]

Both of my filter caps are showing signs of deterioration so I reckon I dodged a bullet there so in this instance, penis fingers was wearing a condom I guess 

[Mr. Scram]

Keeper. GDM-8341 is going however. I either had to pay cal for that or get something better and get it cal'ed. I end up with more cash and more digits at the end of this

Do you actually need a calibration for something or is it a matter of some nice TEA?

[bd139]

Keeper. GDM-8341 is going however. I either had to pay cal for that or get something better and get it cal'ed. I end up with more cash and more digits at the end of this

Do you actually need a calibration for something or is it a matter of some nice TEA?

Good question.

For me, yes. I come here for the TEA but 80% of my “operating time” is forward engineering and everyone owes it to themselves to use the sharpest tools when doing that. This is for the sake of confidence, consistency and communication.

Also I’ve been doing a lot of precision stuff recently (ADC with ridiculous 90dB dynamic range and modulation) and at the extreme edges of measurements is where the calibration makes a difference.

[Cerebus]

Once the cal data on an HP3478A has been nuked its accuracy is the same as the initial accuracy of its internal voltage reference. In the case of the 3478 that's an LM399 with a 5% initial tolerance - it's very stable (0.3 ppm/C, 8ppm/√1000hrs drift) but has a wide "reverse breakdown voltage"* spec of 6.75 – 7.3 V. So until you re-cal it, it's pretty much useless for all but strictly relative measurements.

* It's slightly misleading calling it this as they write the spec sheet as if it's a bare heated zener, when in fact there's a whole circuit pretending to be a zener.

[GerryBags]

Seeing your 3478As, it reminds me that I also have 5 of them - and never even looked at 'em. Instead, I keep buying more stuff (to be shown here soon).

So.... you probably wouldn't miss one if you sent it to me, then? 

[bd139]

Once the cal data on an HP3478A has been nuked its accuracy is the same as the initial accuracy of its internal voltage reference. In the case of the 3478 that's an LM399 with a 5% initial tolerance - it's very stable (0.3 ppm/C, 8ppm/√1000hrs drift) but has a wide "reverse breakdown voltage"* spec of 6.75 – 7.3 V. So until you re-cal it, it's pretty much useless for all but strictly relative measurements.

* It's slightly misleading calling it this as they write the spec sheet as if it's a bare heated zener, when in fact there's a whole circuit pretending to be a zener.

That’s about in line with my observations. I recalibrated the 10v range with my reference. The meter was reading 9.6943 for a 10.0012v input so about 3% out. It also had a zero offset.

[Specmaster]

To be fair you can do that with the 3478A. You just need the reference

Ok, its all put back together and working. I looked at the manual and yes its doable apart the AC side because some of the calibrations require an input of 280v @ 100KHz and there's a few other like it.

I did try to calibrate the 3Vdc range as I a 2.5v voltage reference and it requires the value to input manually and it means holding the arrow keys (up or down) until the display reads the known value precisely and the releases the arrow key and storing that value and it takes bloody ages to get to the value as it counts up 1 at a time from the least significant digit  so I think I'll be getting mine done for me, its a balls ache job and I don't think (may have RTFM for this) it is something that can be done in stages, save it and start again another day on the ranges you haven't yet done, as I said I need the manual to confirm.

Good news is the there is a centre in Brentwood, another in Romford and even 1 in Chelmsford so I may speak to them on Monday and see what they cost. The one in Brentwood can even do it while you wait at no extra cost.

I was going great there as well and then I made the school error of pulling solder from the reel without thinking to insulate the bloody chassis. As I soldered the positive of the back up battery into place, the solder was draped across the frame of the meter which was connected to the negative, in short I fucked it. 

[bd139]

It’s essier if you have a proper calibration generator. If it kicks out 10.000 volts it will read that straight in for esample on each range. Probably a 20 minutes job for a calibration tech. Don’t sweat it though; every mistake we make helps someone else not make one in the future

[Specmaster]

Very true, I managed to avoid the lead clipping accident nicely. Someone's has also fitted a capacitor on the votive ram to retain memory while replacing the battery. Also what ever the battery voltage reads, the volts across the ram is O. 2V  lower. My battery is now 3.2v so I should be a few years before we need to repeat this again.

[Mr. Scram]

That's two for two now. The 3478A turns out to be a bit of landmine. Is something like a 3457A as finicky?

[Specmaster]

That's two for two now. The 3478A turns out to be a bit of landmine. Is something like a 3457A as finicky?

No, I don't think the 3478A is finicky, its just bad luck that bd139 and myself have for what reasons been unlucky. In both cases it was the same thing that done our calibration in, shorting the voltage across the NVRAM causing it to dump the cal data. Its a mistake I bet we never make again.

I think to be honest its a failing of all the more modern meters which have the software driven calibration  that require a battery to retain the data. That battery will at some point need replacing before it fails and accident while replacing could erase the cal data. I would thought it might be better to use something like an EEPROM that does not requires batteries, may be newer meters do?

[Cerebus]

3457A - calibration or replacing the battery?

Calibration is a an order of magnitude more fuss than the the 3478A. You need more and better signal sources and some of the calibration is open case. (The 3478A is all closed case except for a one time factory AC calibration). The calibration section of the manual is at least three times as long, probably more.

Replacing the battery? Probably easier. A bigger case, more room to work. Dave's got photos of the inside somewhere, hold on, I'll see if I can find them. Teardown video thread here: https://www.eevblog.com/forum/blog/eevblog-426-hp-3457a-multimeter-teardown/msg193982/#msg193982. Bags of room around the battery as long as ones does not do anything (* cough *) careless.

[Cerebus]

I think to be honest its a failing of all the more modern meters which have the software driven calibration  that require a battery to retain the data. That battery will at some point need replacing before it fails and accident while replacing could erase the cal data. I would thought it might be better to use something like an EEPROM that does not requires batteries, may be newer meters do?

They do. If you look at the age of the designs that use battery-backed RAM they are from the era where guaranteed writeability of EEPROMs was on the order of 10,000 re-writes, not a figure I'd be happy to bake into the instruments lifetime if I was the designer. Especially when each single calibration (i.e. combination of measurement type and range) counts as using up one of those writes. From memory, a full calibration on the 3478A bangs up the calibration count by 31.

A quick look shows the 3457A written up in the Feb 1986 HP Journal. I'd imagine the design was probably at the parts selection stage two years earlier, so think 1984 or earlier for the EEPROM technology available. The 3478A is actually a slightly older design, the article on the 3457A mentions using the same ADC/switching hybrid in both. Earliest serial number I can find for a 3478A is week 36 1981 (from the updating section of the service manual).