Test Equipment Anonymous (TEA) : Discussion and Group Therapy Thread

[bd139]

The Philips blue ones are absolutely fine.

Until they go short!

[mansaxel]

So, new arrival (talked about it in Discord, but that's only half of the introduction to the community.):

-hp- 3478A Multimeter, my first real foray into "many digits". It was not very expensive, but not as cheap as I would have wanted. Contrary to the cautions made by the very decent seller it seems to work quite OK. The calibration has been lost in a battery swap though, so I will have to amass the "reference" sources required for that. 300V DC seems to be the issue for me here; the rest can be dealt with. A -hp- 6209A would come in handy.


(Seller pic)

I have one issue, and that's what looks like a cold joint on the protection zener for one of the 5V supplies. The regulator regulates, and measuring on the legs of the zener gives that the voltage across is consistent with the 4,98V I measure at the test point; indicating that the joint is indeed one, but any failure will of the regulator can have dire consequences, since there are quite many volts regulated to heat in that 7805, and we do not know how a marginal cold joint will react when being in path of the back-up regulator[0]. Getting to the zener will require disassembly.

The seller also warned about the PCB being discoloured by heat in that same area. I did lower the burden on the supply by moving the primary tap to the "240" leg from "220". This clearly noticeably removed heat from the regulation circuitry. We normally hover around 235VAC here, so "240" clearly is the better choice of those.


(Also seller pic)

[0] The elegance of using a zener diode to limit voltage brings to mind the 1940s solution to keeping the lights at 120V aboard US submarines.

[tggzzz]

One of the joints on that diode is really strange. It looks like some form of ferrule, but doesn't look like it is an "aftermarket bodge". I wonder if it is to allow temporary disconnection by bending the diode out of the way.

Shame about the cal. The 300Vdc into 10Mohm (or lower) isn't trivial, nor is accurate Vac. My trick at lower voltages is to have a DDS function generation, set it to 0.01Hz, capture Vmin and Vmax using a DC voltmeter, and presume those won't change at up to a few kHz.

The only place I have 1kV is in a null voltmeter, driving its internal KVD. I bought an electrophoresis PSU to drive my 100kohm KVD, but it is excessively noisy due to it continually adjusting the voltage and current against my set limits (250W at 1kV? a 10mA current limit isn't a bad thing!)

[bd139]

They are ferrules. It's to stop the clamp diode desoldering itself before the fuse pops

[mansaxel]

They are ferrules. It's to stop the clamp diode desoldering itself before the fuse pops

Oh, really?

Cool, if so.

Anyway, needing 300V DC does happen from time to time, so a supply for it is in planning.

[Zenith]

So, new arrival (talked about it in Discord, but that's only half of the introduction to the community.):

-hp- 3478A Multimeter, my first real foray into "many digits". It was not very expensive, but not as cheap as I would have wanted. Contrary to the cautions made by the very decent seller it seems to work quite OK. The calibration has been lost in a battery swap though, so I will have to amass the "reference" sources required for that. 300V DC seems to be the issue for me here; the rest can be dealt with. A -hp- 6209A would come in handy.

HP 3478As turn up on ebay UK for various prices from about £60 (usually Parts only) to £250 or more, but with no particular assurance they are up to snuff.

I'm starting to get into bench multimeters, but it does occur that it's as well to do your homework on these and profit from the experience of others. Some have EPROMs which have never been copied and some have firmware and hardware which isn't up to scratch and can't be upgraded. Then of course there's the question of calibration, which is either expensive or leads you into volt nuttery.

However, if you can resist the temptation of the quest for 8½ digits, a modest 4½ digit number is a pleasant and useful thing to have. It sits there at the back of the bench, it doesn't use batteries, it doesn't turn itself off in the middle of a lengthy investigation, in most cases the display is easy to read - no peering at dim LCDs, some have GP-IB or other means of data capture and control, which could be useful.

[tggzzz]

So, new arrival (talked about it in Discord, but that's only half of the introduction to the community.):

-hp- 3478A Multimeter, my first real foray into "many digits". It was not very expensive, but not as cheap as I would have wanted. Contrary to the cautions made by the very decent seller it seems to work quite OK. The calibration has been lost in a battery swap though, so I will have to amass the "reference" sources required for that. 300V DC seems to be the issue for me here; the rest can be dealt with. A -hp- 6209A would come in handy.

HP 3478As turn up on ebay UK for various prices from about £60 (usually Parts only) to £250 or more, but with no particular assurance they are up to snuff.

I'm starting to get into bench multimeters, but it does occur that it's as well to do your homework on these and profit from the experience of others. Some have EPROMs which have never been copied and some have firmware and hardware which isn't up to scratch and can't be upgraded. Then of course there's the question of calibration, which is either expensive or leads you into volt nuttery.

However, if you can resist the temptation of the quest for 8½ digits, a modest 4½ digit number is a pleasant and useful thing to have. It sits there at the back of the bench, it doesn't use batteries, it doesn't turn itself off in the middle of a lengthy investigation, in most cases the display is easy to read - no peering at dim LCDs, some have GP-IB or other means of data capture and control, which could be useful.

I like meters that can display more than a single value.

Previously I've used RS232/HPIB to dump and post-process streams of readings, but I do like being able to simultaneously see the reading, min/max/mean/stddev.

In that sense my current favourite is my Agilent 34410A (not 01A), with a VFD display. 6.5 digits, which really is more than enough for anything sensible

No, I'm not "losing" my 7081 nor 7075 not 2015THD nor 2000, nor 1061 nor ....

[mansaxel]

I like meters that can display more than a single value.

I would concur, even if it is from the point of much simpler devices -- my 3 1/2-banger Scopemeter123 does two measurements/traces and also remembers the last unit measured, (with some exceptions, volts after ohms would probably not be sensible, but DCV after ACV most certainly is!) so you will have 4 measurements going, with a fantastically pixly, slow, picture of the waveform or trend.

[Zenith]

Well, you can get carried away with this stuff and on the principle that you can't take it with you when you go, a nice little Siglent SDM3065X Digital Multimeter (or others) might hit this particular ball out of the ground, for all practical purposes.

https://siglent.co.uk/product/siglent-s ... ultimeter/

Nonetheless, buying one for £750 inc VAT, would seem a bit like cheating.

[bd139]

I'd take an Agilent for half that and pay to get it calibrated. Or an HM8112-3

[Zenith]

I'd probably do the same. Still if you had a serious use for it, and could reclaim the VAT, £750 with a guarantee and a calibration certificate is worth a thought.

What does it cost to get a 6½ digit meter calibrated? I've had a quick look and calibration houses are a bit cagey about publishing their prices, you have to apply for a quote. I'd guess £150.

[tggzzz]

I'd probably do the same. Still if you had a serious use for it, and could reclaim the VAT, £750 with a guarantee and a calibration certificate is worth a thought.

What does it cost to get a 6½ digit meter calibrated? I've had a quick look and calibration houses are a bit cagey about publishing their prices, you have to apply for a quote. I'd guess £150.

Most calibration houses are set up to give a piece of paper saying your 3.5 digit meter is within spec.

Getting a piece of paper a "your meter said X and our meter said Y+-uncertainty" is something you have to make sure happens

Adjustments? What do HPAK/Keithely/Fluke charge

[Zenith]

This lot do the Full Monty, 3½,4½,5½ and 6½, voltage,current and resistance UKAS accredited, but you have to apply for a quote.

https://www.rhopointmetrology.co.uk/cal ... ultimeter/

They also do capacitance meters and calibrators "traceable".

I'd guess they adjust meters which are software adjustable, but are less keen on opening things up and twiddling presets, all for extra cost. In their position I'd take anybody's money for a certificate such as you describe, assuming the TE in question gave a stable reading. I might refuse to adjust anything over say 20 years old, on the grounds that too much could go wrong.

HP/Agilent/Keysight, R&S, Fluke, Tek will all offer cal services and repair and possibly repair on a best effort basis for items long off the price list. I bet the prices would make your eyes water.

[Robert]

My recntly aquired Wavetek Model 52 not ony had dual display it has two fully isolated input modules and each of those can measure voltage and curreet and thus calculate power at the same time. All for £35. My Keithley 199 is probably favorite though. No dual display but compact with 51/2 big clear red LED digits.
The stack of in-use bench meters is Fluke 8853A (6.5 digit), Keithley 199 (5.5 digit), R&S UDS5 (5.5 digit) and Wavetek 52 (4.5 Digit x 2).

[Zenith]

The stack of in-use bench meters is Fluke 8853A (6.5 digit), Keithley 199 (5.5 digit), R&S UDS5 (5.5 digit) and Wavetek 52 (4.5 Digit x 2).

Swanking it and rubbing it in. I like it. I'm only just starting with the bench multimeter scene, but I must admit there's something very alluring about it. A world of wonder.

Do tell us about your calibration standards and how you know these aren't conspiring to produce convincing looking rubbish.

[25 CPS]

A whole bunch of LED lights on my truck failed in interesting ways fairly close together. Some went dim, some flickered, and I ended up breaking down and buying four to replace the two above the rear license plate and the front marker lights. That's when I discovered the truck's previous owner must've bought dubious e-commerce website aftermarket garbage. Here's an example of one of these lamps that can't be considered a proper automotive product. This more of a proof of concept or a prototype, at best:







It's the same on both sides of the circuit board, two little surface mount diodes, one 24 ohm resistor that both checked out, and one LED. There's a third LED facing out from the end piece that's soldered on. The 34401As got fired up for continuity and diode tests and all of the parts on the lamp I tested checked out and yet this one failed by going dim in service. I know it's not the truck because the replacement worked fine once I installed it so I did a power test on the bench and this is where it got really interesting:





At about 6 volts, the bulb came alive and was blazing at 8 volts then between 8.5 and 9 volts, the whole thing would go dim - and stay that way until the voltage was turned all the way down and then brought back up to the 6 to 8 volt range that it worked at. Going from 9+ volts back down to 8 volts didn't make any change, it had to be turned all the way down or off and then brought back up. Consequently, the bright picture was taken at 8 volts and the dim one at the full automotive 12 volts. This was repeatable too. Unfortunately, I didn't have time to take any measurements across the LEDs to see what was happening between full brightness and going dim, and in hindsight, I should've looped the power feed through another multimeter to be able to observe current consumption as well. I kept the failed lamps so this is definitely a little project for later because I'm curious how exactly these garbage aftermarket LED lamps are failing.



Also on the subject of lighting, I had a beautiful fall afternoon drive to pick up this General Radio Strobotac. It's going to be fun using this to do motion studies, eg. of speaker driver movement etc.

[Robert]

I've seen this failure mode before. I was working with high current pulse operation of LEDs and seeing how far we could push them. This was on the day job nearly 20 years go. If you ran them too hard the light output whent down along with the forward voltage. If you backed the current off the brightness and Vf incresed but not to the same level. as before the damge. Looking at the LED chip under a microscopr it was clear the some areas wher not producing light. These areas seemd to become some kind of negative resistance with the Vf falling thus shutting down the working part of the chip. Don't know what low level failure mode was but possible migration of doping due to localised heating. Any semiconductor specialists out there care to comment?

[Zenith]

It seems like a plausible explanation. Tunnel LEDs, the component of the future.

These things are operating at the limits.

Here's something I've been curious about. It's the PCB from a 50W work light which failed. It was dead and was thrown out. It produces nothing when 230V is applied so probably not the same failure mode as the auto lights 25CPS has. The unit was cast aluminium and the glass front was glued in place, so it had to be broken to remove the PCB. The PCB is aluminium; it has some sort of insulation with tracks glued to a sheet of aluminium which was in turn glued to the case with some sort of thermal paste/glue. The case has fins on the back.

DSCN2411.JPG

It see some resistors and maybe a couple of over voltage suppressors from the mains to a bridge rectifier. Then some sort of regulator chip working on 230V FW rectified with no reservoir caps, then what is probably an IGFET and then an array of LEDs. The regulator chip and IGFET are made by Reactor Microelectronics, a Chinese company which makes automotive and power control devices. They have a website and there are brief datasheets/specsheets available in Chinese. The devices in this LED unit are not covered on their website. Nothing is obviously burned or damaged.

I had hoped that BigCliveDotCom on YT would have investigated this or something similar, as he looks into a lot of Cheapanese electronics, but no.

It's interesting to look at, but I really can't be bothered to investigate it further and I'll throw it away in due course. What I will say is if you are going to operate one of these units, make damned sure the body is earthed.

[AVGresponding]

If none of the MELF's are melted, I'd suspect an o/c on one of the NTC's (or are they PTC's, some sort of primitive pre-regulator? It's not 100% clear) or a popped transistor. If one of those MLCC's had failed, you'd probably be getting a flash of some sort when you powered it. There's no inductors visible, so it's going to be a linear regulator.