KX Reference

[TiN]

Looking better now. Should be no much difference running from battery or DC supply, if all grounding done right.
I'd avoid any kind of clips or connectors between the reference and meter and use freshly cut copper twister pair (CAT5e/6e lan cable works best for that, with shield connected only one side, at DMM guard LO).

[Muxr]

Was about to order a LTZ1000A.. digikey is out of stock, ordered it directly from LT but I also got a warning that it will likely take a while. Which one of you bought up all the LTZ1000As? 

TME in europe has them but they are a bit more expensive. Will just wait until LT ships me mine.

[TiN]

That warning is there as long as I remember, but never was a problem for few pcs qty I was ordering before.

[kj7e]

Quick update on my KX portable battery powered ovenized enclosure.  Everything fits and works, just waiting on the Vishay resistors before I can complete the buffer.

https://youtu.be/OuonZGQq0Q4

[TiN]

I like it. What is the case model? It seems perfect size for such kind of project. Next step to put some Raspberry Pi inside for datastorage/logging and external meter control, so you can always keep digital history of the reference together with the box, making it perfect travel standard.

[kj7e]

I like it. What is the case model? It seems perfect size for such kind of project. Next step to put some Raspberry Pi inside for datastorage/logging and external meter control, so you can always keep digital history of the reference together with the box, making it perfect travel standard.

Bud Industries TR-6100;
http://www.budind.com/view/Small+Metal+Electronics+Enclosures/TR+Series+Small+Cabinet

If you search around you they can be found at a discount.  The Die Cast box used for the oven/shielding is a Bud Industries CU-4234.  The Raspberry Pi with environmental logging is a good idea, maybe for V2.  Once this one is built, I hope to send it off to someone with a freshly calibrated 3458A or better to measure/adjust.

[lukier]

kj7e: nice build. Thanks for the info on the case. What are you planning for the output protection (short circuit, feeding in voltage, maybe gas discharge tube)?

[Edwin G. Pettis]

The Bud case is very nice and very well made, also expensive.  I used a TR6102 for my winding machine controller and the last time I checked they had jumped to $225.00 (about $100 higher than they used to be).  Yipes, I just checked latest pricing and they're up to $275.00, ridiculous.  I did find a 'discounted' price of $244 at one place, still pricey.

[kj7e]

The Bud case is very nice and very well made, also expensive.  I used a TR6102 for my winding machine controller and the last time I checked they had jumped to $225.00 (about $100 higher than they used to be).  Yipes, I just checked latest pricing and they're up to $275.00, ridiculous.  I did find a 'discounted' price of $244 at one place, still pricey.

They can be found here for $174;
http://www.aiconsol.com/tr-6100-bud.html

Still not cheap, but I wanted to make this a nice piece of equipment I could admire for years, and I'm having a blast building and learning.

[kj7e]

kj7e: nice build. Thanks for the info on the case. What are you planning for the output protection (short circuit, feeding in voltage, maybe gas discharge tube)?

The output buffer is going to be the LTC1150/LT1010 combo. The LT1010 is fairly robust and can withstand shorts and other abuse;
http://cds.linear.com/docs/en/datasheet/1010fe.pdf

[SvanGool]

@KJ7E:

Nice design !

If you want this to be an "easy" travelling standard, without the need of guiding "dangerous goods" documents/labels (and the risk that your post/freight company won't take it as a standard package), one cell needs to be below 20 Wh (that was the easy part) and your total battery (consisting of one or more cells) capacity needs to be below 100 Wh (that is the difficult part) and you cannot have more than 4 cells or two batteries, according IATA Battery guidance 2017 for air freight UN3481 PI (Package Instruction) 967 Section II . If you are able to accommodate that, then that would make your solution fly. 

In another project, I was looking for 8x (4S2P) Samsung INR18650-35E (12.24 Wh according official Samsung specs), which just fits the bill.

I am curious, can you give some more detail on your oven control circuitry?

[kj7e]

@KJ7E:

Nice design !

If you want this to be an "easy" travelling standard, without the need of guiding "dangerous goods" documents/labels (and the risk that your post/freight company won't take it as a standard package), one cell needs to be below 20 Wh (that was the easy part) and your total battery (consisting of one or more cells) capacity needs to be below 100 Wh (that is the difficult part) and you cannot have more than 4 cells or two batteries, according IATA Battery guidance 2017 for air freight UN3481 PI (Package Instruction) 967 Section II . If you are able to accommodate that, then that would make your solution fly. 

In another project, I was looking for 8x (4S2P) Samsung INR18650-35E (12.24 Wh according official Samsung specs), which just fits the bill.

I am curious, can you give some more detail on your oven control circuitry?

Not too worried about shipping here in the US, just have to send it UPS or FEDEX ground.  I wanted the capacity and the LiPo's where inexpensive (Hobby King).

Back a page, I posted some more info on the oven control circuit, here is the post;
https://www.eevblog.com/forum/metrology/kx-reference/msg1241349/#msg1241349
I have to admit, it works better than I could have hoped for.

[Edwin G. Pettis]

KJ7E,

I checked out the link, the $174 is for the 6100 and they don't seem to have the TR-6102 listed.  Back in the day when I bought this unit for the controller, it was a 'paltry' $125.00 and that was expensive back in the latter '90s.  The TR-6102 actually cost more than the PCB and all the components on it, the PCB is about 12.5" L x 7" W.  Now, that case is close to >2.5 times the components.

[MisterDiodes]

KJ7E,
Head's up using LT1150 / '1010 on a Vref.  Watch your noise density specs on the '1010 especially and the '1150 is a pretty slow & noisy chopper compared to newer designs - those parts are ancient history these days and will splatter noise downstream if you're careful.  I would use a simpler, cheaper 10-cent transistor follower output drive circuit over a '1010 for much lower noise - IF you don't need push-pull output current.  It just depends on what you'll use this for.

I suggest you test the output circuit very carefully to make sure it's what you need and matches up to what you can measure (or not).

[kj7e]

KJ7E,
Head's up using LT1150 / '1010 on a Vref.  Watch your noise density specs on the '1010 especially and the '1150 is a pretty slow & noisy chopper compared to newer designs - those parts are ancient history these days and will splatter noise downstream if you're careful.  I would use a simpler, cheaper 10-cent transistor follower output drive circuit over a '1010 for much lower noise - IF you don't need push-pull output current.  It just depends on what you'll use this for.

I suggest you test the output circuit very carefully to make sure it's what you need and matches up to what you can measure (or not).

I'm open to suggestions,  without derailing this thread too much, what would you recommend over the 1150?  I was just basing this on LT App note 86 page 46.

[The Soulman]

It appears to me that we have some very knowledgeable people here, is it possible for them to write up a recipe with
today's best parts and practices?
With so commentary on how and if and such.
And perhaps make that a sticky so they won't have to answer these questions every couple of weeks. 

[lukier]

I guess the answers are probably in the LTZ1000 thread

[hwj-d]

KJ7E,
Head's up using LT1150 / '1010 on a Vref.  Watch your noise density specs on the '1010 especially and the '1150 is a pretty slow & noisy chopper compared to newer designs - those parts are ancient history these days and will splatter noise downstream if you're careful.  I would use a simpler, cheaper 10-cent transistor follower output drive circuit over a '1010 for much lower noise - IF you don't need push-pull output current.  It just depends on what you'll use this for.

I suggest you test the output circuit very carefully to make sure it's what you need and matches up to what you can measure (or not).

I'm open to suggestions,  without derailing this thread too much, what would you recommend over the 1150?  I was just basing this on LT App note 86 page 46.

Because i'm learning this stuff at the moment, that's a very interesting question to me too. I also orient myself to this old but very informative documents (Jan 2001). Are there comparable practice-oriented Infos like this for newer designs with modern Op Amps out in the net?

OT: Agree with The Soulman, it would really help- and thankful, if someone could explain in short these differences to modern chips on the market. Specially in relation to use them with references.

[Muxr]

I guess the answers are probably in the LTZ1000 thread

From my reading at least as far as LTZ1000 is concerned, there are two schools of thought, when it comes to which Op Amps to pair LTZ1000 with.

- modern chopper LTC2057 (this is what TiN's KX uses)

- or the conventional LT1006, LT1013 (dual) which LTZ1000 specifies in the datasheet and Linear recommends.

[lukier]

From my reading at least as far as LTZ1000 is concerned, there are two schools of thought, when it comes to which Op Amps to pair LTZ1000 with.

- modern chopper LTC2057 (this is what TiN's KX uses)
- or the conventional LT1006, LT1013 (dual) which LTZ1000 specifies in the datasheet and Linear recommends.

Well that's just the opamp for the LTZ1000 circuit itself and I must say MisterDiodes' arguments (from some other thread) against LTC2057 seem sensible.

Here, especially in case of a nice box-standard build, like KJ7E does (not just the PCB with LTZ1000 page 6 circuit) I would be interested in the output buffer and the related protections. Frying LTZ1000 is something I would really like to avoid

Fluke 732B schematics offer some hints. They seem to have some diode and TVS across the output along with some capacitors, gas discharge tube from LO to earth and something that looks like transistor short circuit protection (Q1,C1,Q2,R1) and voltage clamp (CR1), see A2 flex circuit PCA and A1 front panel PCA schematics.

[hwj-d]

I would be interested in the output buffer and the related protections.

That's the point.

[TiN]

Meanwhile I'm getting VK5RC's KX reference ready for shipment back to Oz cal-club.

Before





Tempco is bad, expected below 0.05ppm/K, typical to LTZ ref.

After

I didn't quite like the way board was mounted, so I got it free hanging, surrounded by foam. Then rewired copper junction from zener opamp to match B03 PCB version (VK5RC used older B01 in his build). And dropped LTZ package flush to the board, instead of standing on the legs. As result output shifted -15ppm.

Note different PPM scale +1/-1ppm instead of 0/-3ppm on first graph.



Yay, tempco is beyond measurement ability of K2002 over 17C deg span. As expected.

And since no thread is good without photos, here they are:



CSV data as usual Run 1, Run 2, started Run 3 with slower ramp.

Now just need to get final calibration using 4 of my meters, and it's finally ready to go to VK5RC next monday. 

[VK5RC]

Thanks for your efforts TiN,
Interesting results, I take from it,
1 Mr Pettis makes nice resistors, esp in value compared to Vishay,
2 The 2057 does quite well,
3 Is the LTZ 1000 especially vulnerable to air currents from below? 
4 The changes of Kx board  from B01 to B03 important.
5 You can mount wire wound type resistors in the vertical fashion without too much compromise.
6 The mounting of the Kx board is best by foam suspension. - I need to take the covers off the 3458 and see how HP mounted their LTZ. I recall? 2 lots of pin headers, but poss other mechanical fixtures .

Again thanks, once back, I will look at how far off my 3458 is, and then forward it around the Aus Cal Group.
Regards Rob 

[julian1]

I like it. What is the case model? It seems perfect size for such kind of project. Next step to put some Raspberry Pi inside for datastorage/logging and external meter control, so you can always keep digital history of the reference together with the box, making it perfect travel standard.

Bud Industries TR-6100;
http://www.budind.com/view/Small+Metal+Electronics+Enclosures/TR+Series+Small+Cabinet

If you search around you they can be found at a discount.  The Die Cast box used for the oven/shielding is a Bud Industries CU-4234.  The Raspberry Pi with environmental logging is a good idea, maybe for V2.  Once this one is built, I hope to send it off to someone with a freshly calibrated 3458A or better to measure/adjust.

Is the carrying handle/tilting bail aluminium or plastic? It looks like it might be a bit fragile if plastic.

[kj7e]

Is the carrying handle/tilting bail aluminium or plastic? It looks like it might be a bit fragile if plastic.

Aluminum, with a neoprene wrap.  Includes Teflon washers and nice three prong adjustment knobs as well.