Test Equipment Anonymous (TEA) group therapy thread

[ChrisLX200]

That looks similar to mine. I read somewhere (lost the source) that it needs to be powered up for ~20-30 mins before it's stable.

[Specmaster]

Yes, I read that somewhere, but then, they say about everything don't they! I really doubt that for the average person on here its really worth it just to be accurate by maybe 1 or 2 millivolts, its hardly for the best part going to make that much difference on but the most delicate of circuits is it.

[HalFET]

Yes, I read that somewhere, but then, they say about everything don't they! I really doubt that for the average person on here its really worth it just to be accurate by maybe 1 or 2 millivolts, its hardly for the best part going to make that much difference on but the most delicate of circuits is it.

I must say I often run into microvolts for my home builds, then again it might depend on what you're building.

[bd139]

Only place I find microvolts is op amp offset trimming and that’s just professional pride as I’m not doing any precision work.  And you don’t need a calibrated meter for that as it’s a differential measurement. As long as your meter zeroes when shorted.

Going to be honest here and I’m usually good with a 5% reading and won’t be buying anything better than a 4.5 digit meter. Not because I don’t want one but I can’t calibrate one that accurate so it’d sit in my mind and annoy me.

[Specmaster]

That used to be me, 5% was pretty accuracy for most things I thought, but then it used to bug me when working on something to day and it measures x and then the next day I'd maybe grab a different meter and it'd read y so thats why I grabbed a reference and then adjust meter to be as close to each other as I possibly get them.

[HalFET]

Only place I find microvolts is op amp offset trimming and that’s just professional pride as I’m not doing any precision work.  And you don’t need a calibrated meter for that as it’s a differential measurement. As long as your meter zeroes when shorted.

Going to be honest here and I’m usually good with a 5% reading and won’t be buying anything better than a 4.5 digit meter. Not because I don’t want one but I can’t calibrate one that accurate so it’d sit in my mind and annoy me.

Well, I'm designing my own 6.5/7.5 digit multimeter at the moment, so that might explain why. I'm pretty much in need of an electrometer at this point since the charge injection is pretty annoying to estimate, so might just measure it directly instead    I feel an excuse coming up!

[URI]

I have never seen a post with plastic threads on them, even some cheap one hung low ones I purchased use a metal insert that is threaded on ends so the plastic cover nut and the metal retaining nuts, all screw onto the same "bolt" hence they attach securely.

Ok, her you are, the BS-244 series Binding Post with plastic thread:


The ones I showed with my front plate (bought them from pollin electronic) are built as you described:


Either one has downsides.
The BS-244 series has a plastic thread for mounting it on front plates.
The pollin binding post is built from a piece of metal that is threaded and the plastic cover nut and the metal retaining nuts all screw onto the same "bolt" as you mentioned. Fastening is easy and reliable.
I decided to stick with the BS-244 series because of a downside of the pollin binding posts that's decisive for me: How deep can I push a plug in?
The BS-244 series starts to grip the plugs contact spring much earlier because the isolation cover is smaller:


The Pollin binding post starts to grip the plugs contact spring later and therefore it can't be pushed in as far as into the BS-244 series binding post. Shorter plugs with spring loaded safety shutter only have grip for 2mm pushed in and therefore fly out often when the cable and with it the plug is moved slightly:


Another downside of the pollin binding post is the cap that can be fully removed but the cap of the BS-244 series binding post is built as a captive one.

[bd139]

That used to be me, 5% was pretty accuracy for most things I thought, but then it used to bug me when working on something to day and it measures x and then the next day I'd maybe grab a different meter and it'd read y so thats why I grabbed a reference and then adjust meter to be as close to each other as I possibly get them.

As long as they all match that’s most important.

I’m treating my Keysight as the “source of truth” as that has an August 2017 cal cert. Everything matches so I’m happy

Only place I find microvolts is op amp offset trimming and that’s just professional pride as I’m not doing any precision work.  And you don’t need a calibrated meter for that as it’s a differential measurement. As long as your meter zeroes when shorted.

Going to be honest here and I’m usually good with a 5% reading and won’t be buying anything better than a 4.5 digit meter. Not because I don’t want one but I can’t calibrate one that accurate so it’d sit in my mind and annoy me.

Well, I'm designing my own 6.5/7.5 digit multimeter at the moment, so that might explain why. I'm pretty much in need of an electrometer at this point since the charge injection is pretty annoying to estimate, so might just measure it directly instead    I feel an excuse coming up!

I can understand it for that!  Had some problems with that myself and solved it by reducing impedances. Lazy but effective. No good for a DVM front end though!

[URI]

Yay new power supply arrived from SGLabs.it. comes with a printed manual and CD and UK mains cable and works perfectly. Good job

Very nice piece of kit!
And -ahhh- it's marked "hp" that's a symbolic plus on top.
Don't really like the agilent nor the keysight logo.. 

[bd139]

Yes I much prefer to see HP written on it

[URI]

Hmmm... I think I see your problem better now; particularly the issue of weak plastic threads. Either you need other tools as I suggested or to look for more suitable, but similar binding posts.

You are right, I don't think I have the ability or the tools to make my own special tools.

Okay... how about this: You have a mini drill press in your setup; so repetitive small drillings should be no problem. I suggest making backing plate(s) of ~1-2mm metal (such aluminum is readily available from any Homeowner Hell type hardware store in their specialty bins), which you can hand-fab the D-shaped holes into pretty easily even with your basic tool set. Make the plates it so they span at least two posts at a time, which by nature will anchor them against applied torque.

This will allow you to preserve the look of your pristine aluminum extrusion, while still providing the torque resistance you desire.

That's exactly what I tried to explain.
I tried my very best but English is not my mother tongue. 
In fact I had the idea to build one filling plate with holes for all binding posts in one plate. I can use my front plate as a model to transfer the position and diameter onto that filling plate.

I've made similar plates where I fold the metal to make the flat for more contact area against the torsional force, but that requires that you allow for the thickness of the metal when drilling/cutting the basic shape and is a hand-fab skill you need some experience to manage.

Thank you for your hints but I think I'll stick with the easier way anyway.
Don't want to do uncounted bad plates to gain the required experience for three plates in the end. 

[Specmaster]

I decided to stick with the BS-244 series because of a downside of the pollin binding posts that's decisive for me: How deep can I push a plug in?
The BS-244 series starts to grip the plugs contact spring much earlier because the isolation cover is smaller:


The Pollin binding post starts to grip the plugs contact spring later and therefore it can't be pushed in as far as into the BS-244 series binding post. Shorter plugs with spring loaded safety shutter only have grip for 2mm pushed in and therefore fly out often when the cable and with it the plug is moved slightly:


Another downside of the pollin binding post is the cap that can be fully removed but the cap of the BS-244 series binding post is built as a captive one.

On the depth of tunnel / tube, I agree with you as thats always a sticking point with me. But to be honest, there is very little between them both, I'd be more inclined to look for better quality plugs that had pins that could be fully inserted leaving nothing exposed and also ones without the 2 extra holes in the body as well. I hate those with the balloon cage around the pin as these never fill me with confidence of getting a reliable connection in the socket.

As to the tops being removable, I'm fine with that as sometimes I want to use clip croc clips on the terminals and I'd rather have 2 points of contact then just 1, plus some crocs won't fit inside the tops.

[HalFET]

That used to be me, 5% was pretty accuracy for most things I thought, but then it used to bug me when working on something to day and it measures x and then the next day I'd maybe grab a different meter and it'd read y so thats why I grabbed a reference and then adjust meter to be as close to each other as I possibly get them.

As long as they all match that’s most important.

I’m treating my Keysight as the “source of truth” as that has an August 2017 cal cert. Everything matches so I’m happy

Heheh, same trick here. I've been using my Keithley 2001 as standard, but I must say it needs to heat up for at least 2 hours before it's rock solid, otherwise you still see some drift on the lower ranges as the reference gets up to temperature. For 10 MHz frequency reference the TXCO in a R&S SMH is a pretty decent one I found. I suppose many of us on here are actually doing it like that? I must also admit I've used instruments at work to do a user-cal on my own gear after-hours.

I can understand it for that!  Had some problems with that myself and solved it by reducing impedances. Lazy but effective. No good for a DVM front end though!

My current trick is attempting to drain the charge elsewhere, but I think I'll just let it flow (charge = spice  ) and take it into account in the FPGA or MCU instead.

Currently aiming for the 1 GOhm input impedance mark, but trying to keep the BOM cost down. I'm fed up with these instruments costing so much, there's no reason for it other than design cost. My first design probably would have reached most of the required specs but would have ended up costing 300-500 EUR in exotic parts + 8 layer PCB. But got rid of the ceramic resistor array, switched to array resistors instead, dumped the fancy op amps, replaced the high-end FPGA with a run-of-the-mill FPGA board from aliexpress. I'm trying to squeeze a stable 6.5 digit into < 100 EUR, basically the main cost at this point is the reference (LM399). There's no real way to save on that one, but I'm trying to cheapass my way out of the ADC now.    If I get to the point where I'm sufficiently happy about the circuit I'll post it up for critical review I suppose, hoping to corner the market of these crappy Chinese bench meters with a decent design with absolutely no intention to make any profit on it. 

[Specmaster]

URI, your English is just fine, its a trillion times better than my German French and thats something you can take to the bank 

[Specmaster]

That used to be me, 5% was pretty accuracy for most things I thought, but then it used to bug me when working on something to day and it measures x and then the next day I'd maybe grab a different meter and it'd read y so thats why I grabbed a reference and then adjust meter to be as close to each other as I possibly get them.

As long as they all match that’s most important.

I’m treating my Keysight as the “source of truth” as that has an August 2017 cal cert. Everything matches so I’m happy

Heheh, same trick here. I've been using my Keithley 2001 as standard, but I must say it needs to heat up for at least 2 hours before it's rock solid, otherwise you still see some drift on the lower ranges as the reference gets up to temperature. For 10 MHz frequency reference the TXCO in a R&S SMH is a pretty decent one I found. I suppose many of us on here are actually doing it like that? I must also admit I've used instruments at work to do a user-cal on my own gear after-hours.

I can understand it for that!  Had some problems with that myself and solved it by reducing impedances. Lazy but effective. No good for a DVM front end though!

My current trick is attempting to drain the charge elsewhere, but I think I'll just let it flow (charge = spice  ) and take it into account in the FPGA or MCU instead.

Currently aiming for the 1 GOhm input impedance mark, but trying to keep the BOM cost down. I'm fed up with these instruments costing so much, there's no reason for it other than design cost. My first design probably would have reached most of the required specs but would have ended up costing 300-500 EUR in exotic parts + 8 layer PCB. But got rid of the ceramic resistor array, switched to array resistors instead, dumped the fancy op amps, replaced the high-end FPGA with a run-of-the-mill FPGA board from aliexpress. I'm trying to squeeze a stable 6.5 digit into < 100 EUR, basically the main cost at this point is the reference (LM399). There's no real way to save on that one, but I'm trying to cheapass my way out of the ADC now.    If I get to the point where I'm sufficiently happy about the circuit I'll post it up for critical review I suppose, hoping to corner the market of these crappy Chinese bench meters with a decent design with absolutely no intention to make any profit on it. 

Yep, thats why I brought a chinese volt ref, cheap enough to afford and its then something I set up my meters to agree with it as much as is humanly possible, spent the afternoon bringing my worst meter into alignment, (ISO-TECH IDM203 a RS meter) and its now within an acceptable tolerance band so now I know that I can use what ever meter comes to hand and its going almost identical in readings, worst case is .004v difference on the 2.5v setting which is way way better then 5%, its actually 0.1198561%, on a meter thats had a hard life is pretty good by my books.

I'm very interested in your meter project, sounds really promising. If or rather when you pull it off, are you planning on selling these or even getting someone to produce them and sell on the open market at all? I could well be interested in one of them.

[bd139]

That used to be me, 5% was pretty accuracy for most things I thought, but then it used to bug me when working on something to day and it measures x and then the next day I'd maybe grab a different meter and it'd read y so thats why I grabbed a reference and then adjust meter to be as close to each other as I possibly get them.

As long as they all match that’s most important.

I’m treating my Keysight as the “source of truth” as that has an August 2017 cal cert. Everything matches so I’m happy

Heheh, same trick here. I've been using my Keithley 2001 as standard, but I must say it needs to heat up for at least 2 hours before it's rock solid, otherwise you still see some drift on the lower ranges as the reference gets up to temperature. For 10 MHz frequency reference the TXCO in a R&S SMH is a pretty decent one I found. I suppose many of us on here are actually doing it like that? I must also admit I've used instruments at work to do a user-cal on my own gear after-hours.

I can understand it for that!  Had some problems with that myself and solved it by reducing impedances. Lazy but effective. No good for a DVM front end though!

My current trick is attempting to drain the charge elsewhere, but I think I'll just let it flow (charge = spice  ) and take it into account in the FPGA or MCU instead.

Currently aiming for the 1 GOhm input impedance mark, but trying to keep the BOM cost down. I'm fed up with these instruments costing so much, there's no reason for it other than design cost. My first design probably would have reached most of the required specs but would have ended up costing 300-500 EUR in exotic parts + 8 layer PCB. But got rid of the ceramic resistor array, switched to array resistors instead, dumped the fancy op amps, replaced the high-end FPGA with a run-of-the-mill FPGA board from aliexpress. I'm trying to squeeze a stable 6.5 digit into < 100 EUR, basically the main cost at this point is the reference (LM399). There's no real way to save on that one, but I'm trying to cheapass my way out of the ADC now.    If I get to the point where I'm sufficiently happy about the circuit I'll post it up for critical review I suppose, hoping to corner the market of these crappy Chinese bench meters with a decent design with absolutely no intention to make any profit on it. 

Letting the charge flow was how they did it with the Hp427A voltmeters. The FETs were leaky of held charge  and temperature dependent so they just poked some current in to offset via a diode. Sneaky design. No op amps or anything fancy!

[HalFET]

Yep, thats why I brought a chinese volt ref, cheap enough to afford and its then something I set up my meters to agree with it as much as is humanly possible, spent the afternoon bringing my worst meter into alignment, (ISO-TECH IDM203 a RS meter) and its now within an acceptable tolerance band so now I know that I use what ever meter comes to hand and its going almost identical in readings, worst case is .004v difference on the 2.5v setting which is way way better then 5%, its actually 0.1198561%, on a meter thats had a hard life is pretty good by my books.

I've often been under the impression that meters need to age to become stable anyway, wouldn't be amazed if your Chinese reference is experiencing something similar. Not sure if it's true or not, but it'd certainly explain some of the things I've run into.

I'm very interested in your meter project, sounds really promising. If or rather when you pull it off, are you planning on selling these or even getting someone to produce them and sell on the open market at all? I could well be interested in one of them.

I'm more worried about the "if" factor, between working on a Ph.D. (in the writing phase at the moment), and attempting to have a social life there isn't much time left for circuit designing/prototyping. But I'd just put the schematic and design files online, totally not interested in making money from it. (Already happy enough with what the university pays me as researcher to be honest, more worried about those dreaded contract extensions. )

But the entire issue with such a thing is that it's most definitely not an average home project to assemble, the high impedances mean you need a very clean board to begin with. Half the prototyping is hovering above a copper plate with the occasional lump of teflon to support things, which ain't ideal either. Currently in a quest to avoid using the AD704 to bias and buffer the reference, it's looking more and more like a discrete opamp frontend job.

[bd139]

Alcohol solves everything. Social life, PhD frustrations, cleaning high impedance traces

[mnementh]

The one I got was a AD854KH which is the more accurate of the chips and as is normal for chinese stuff it arrived with no instructions and the hole in the side where the micro usb plug goes to charge it up was not in the right place so I had dismantle it to enlarge the hole with a file, otherwise it seems fine.

That's the KKMoon Voltage Ref (or a clone); I read about it on EEVblog last year and ordered one from AliEx... it became one of the few transactions I had to file a dispute on over several years of purchases. After that headache  I sortof just let it slide until I stumbled across Mark Hennessy's article a few weeks ago.

I liked the convenience of use, but my long-term experience with LiPos tells me they're a poor choice for a tool that may sit on the shelf for 6 months between uses. That and the fact of using a noisy boost regulator for the power source kindof turned me off on it; I'm a true believer in the simpler the better for such things.

As for which AD854 variant is best, I'm just looking at the chart from the article I linked to; I'll admit I didn't look up the datasheets myself, but it certainly seems to indicate that the LH is considerably tighter tolerance than either of the other two common variants (particularly in temperature stability), and I have no reason to believe Hennessy would misrepresent. Of course, as he also suggests, which variant will actually arrive in the post is probably a crapshoot.

That looks similar to mine. I read somewhere (lost the source) that it needs to be powered up for ~20-30 mins before it's stable.

Yeah, this was one of the reasons I wasn't too sanguine about the whole built-in LiPo thing; I can see letting it run for 30 minutes before using it with that tiny LiPo powering a boost regulator becoming a "Will it run long enough for me to get my work done?" scenario. Most of my 4S RC LiPos are 1000-6000mAH; they'll run that AD854 for days powered directly. 

There was some discussion about whether the likelihood of these being scavenged parts means they come "pre-aged" or not. I'm of the opinion that at $US8, it's a gamble even I'm willing to take.

Only place I find microvolts is op amp offset trimming and that’s just professional pride as I’m not doing any precision work.  And you don’t need a calibrated meter for that as it’s a differential measurement. As long as your meter zeroes when shorted.

Going to be honest here and I’m usually good with a 5% reading and won’t be buying anything better than a 4.5 digit meter. Not because I don’t want one but I can’t calibrate one that accurate so it’d sit in my mind and annoy me.

I agree with you on this 100%, particularly about the "niggling in the back of your mind" part.

Hmmm... I think I see your problem better now; particularly the issue of weak plastic threads. Either you need other tools as I suggested or to look for more suitable, but similar binding posts.

You are right, I don't think I have the ability or the tools to make my own special tools.

Okay... how about this: You have a mini drill press in your setup; so repetitive small drillings should be no problem. I suggest making backing plate(s) of ~1-2mm metal (such aluminum is readily available from any Homeowner Hell type hardware store in their specialty bins), which you can hand-fab the D-shaped holes into pretty easily even with your basic tool set. Make the plates it so they span at least two posts at a time, which by nature will anchor them against applied torque.

This will allow you to preserve the look of your pristine aluminum extrusion, while still providing the torque resistance you desire.

That's exactly what I tried to explain.
I tried my very best but English is not my mother tongue. 
In fact I had the idea to build one filling plate with holes for all binding posts in one plate. I can use my front plate as a model to transfer the position and diameter onto that filling plate.

I've made similar plates where I fold the metal to make the flat for more contact area against the torsional force, but that requires that you allow for the thickness of the metal when drilling/cutting the basic shape and is a hand-fab skill you need some experience to manage.

Thank you for your hints but I think I'll stick with the easier way anyway.
Don't want to do uncounted bad plates to gain the required experience for three plates in the end. 

Oh, no problem... my confusion was where you said "Filler plate", which may just be a language barrier thing. To me that means a plate that goes on the OUTSIDE of a panel to cover an existing hole or a void space in the wrong place; these are a "cosmetic piece" which requires some amount of care in fabrication.

My thought was to make the plates so they go behind everything, so you could just drill your panel out to the oversized holes for the binding posts and then the plates wouldn't be visible. Then they can be as ugly or as tidy as you have the patience to make them. I think you understood me right, so it's all good.

Like SpecMaster said; your English is just fine, its a trillion times better than my German. I know just enough to get my face slapped.


mnem
Confucius say: "A man with a watch always knows what time it is; a man with two is never sure."

[HalFET]

Alcohol solves everything. Social life, PhD frustrations, cleaning high impedance traces

Yes, but can't use vodka to clean circuit boards 

[Mr. Scram]

Yes, but can't use vodka to clean circuit boards 

Sure you can. Will it work well and get the board clean? Eh. Details.

[bd139]

It’s distilled water and ethanol. Perfect

[Mr. Scram]

It’s distilled water and ethanol. Perfect

"Assume a sperical cow"

[HalFET]

Yes, but can't use vodka to clean circuit boards 

Sure you can. Will it work well and get the board clean? Eh. Details.

Yes, but if I use it to clean the boards I can't drink it 

Also this arrived today: https://www.amazon.de/gp/product/B003WGLHD8/

Also, I like my cows spherical, makes rolling them onto the field a lot easier!

[Specmaster]

Yay new power supply arrived from SGLabs.it. comes with a printed manual and CD and UK mains cable and works perfectly. Good job

Just how do you get the photos embedded instead of an attachment? I did manage it once but that was by accident I think as I have never been able replicate it again