[FT] xDevs.com KX LTZ1000A fairy-tale, or the story of little jumper.

[mimmus78]

Hi TiN,

do you still have some of this PCB unpopulated?

Thanks,

Domenico

[TiN]

Hi,

I have two left, but I need them for few items in near future. Somebody here is waiting for fab using my gerbers, I believe.
Btw, I done log with K2001 for you, but have no time recently to make pretty graph. Live graph and RAW data is here. Plenty of data there to run STDEV, noise pk-pk analyze and such. K2001 column is k10v.

[plesa]

Hi,

I have two left, but I need them for few items in near future. Somebody here is waiting for fab using my gerbers, I believe.
Btw, I done log with K2001 for you, but have no time recently to make pretty graph. Live graph and RAW data is here. Plenty of data there to run STDEV, noise pk-pk analyze and such. K2001 column is k10v.

I have 4 TiN boards, but all populated. All are currently measured with recently calibrated 3458A/002.

[TiN]

plesa
How's your datalogging setup going on? If you wish, I can give you space on my serv and script, so your Pi can upload CSVs periodically and I can have live plot page for you

[plesa]

plesa
How's your datalogging setup going on? If you wish, I can give you space on my serv and script, so your Pi can upload CSVs periodically and I can have live plot page for you

Thanks I already have local website on RPi2 which use also the https://d3js.org/ for graphs.
It is quite amazing, BME280 will be connected soon for environment logging.

[TiN]

I'd love to see some data too 

[mimmus78]

Thanks TiN ... noise is identical.

Anyway just ordered 20 pcb of your project, if someone needs some I have few spares.

[pelule]

@ mimmus78

Anyway just ordered 20 pcb of your project, if someone needs some I have few spares.

Just to be sure, I have the right "project" in mind. Does you have ordered this?
https://xdevs.com/doc/xDevs.com/KX_STR/top.jpg

If so, I would be interested into 2 to 4 boards.

[mimmus78]

Well I ordered 20 PCB. I will keep 3 for me and the rest I can sell. I will actually build only one kit, I don't think I will use more than 3 pcb because this kit will cost more than 300 EUR.

I spent 160 EUR per the 20 pcb included express shipping and slow fabrication.  So it's 8 EUR + shipping for one.
Shipping for PCB will be done by registered mail. Usually it takes 10 days for US and one week for europe.
I will not do DHL/UPS shipping because this will require customs declaration and invoicing that I really don't like to do.

I asked for a quote to vishay for restors just for myself... but  maybe we can group buy also this resistors (here is 60%~ of kit costs).
Unfortunately shipping those 200 EUR worth of resistor cannot be done by standard mail ... so it may turn out not so convenient.

As in the other thread, if you like to have some pcb just add your name to the list.

Mimmus78: 3pcs
Nuno_pt: 3pcs
msliva: 3pcs
pelule: 3pcs
==================================
Remaining Boards: 8pcs

[Nuno_pt]

There is no need for the Vishay VHP, you can use UXB, UPW20/50 or from Edwin, 0.1% 3ppm/C°

[klaus11]

Así que pedí 20 PCB. Voy a seguir 3 para mí y para el resto puedo vender. De hecho, me construiré un solo kit, no creo que voy a utilizar más de 3 PCB ya que este kit tendrá un costo de más de 300 euros.

Pasé 160 euros por el 20 de PCB, con el envío expreso y fabricación lento. Por lo que es de 8 euros + gastos de envío para uno.
Envío de PCB se hará por correo certificado. Por lo general, se tarda 10 días en los Estados Unidos y una semana para Europa.
No haré el envío de DHL / UPS, porque esto requerirá declaración de aduanas y la facturación que realmente no me gusta hacer.

Pedí una cita para Vishay para restors sólo para a mí mismo ... pero tal vez podemos agrupar comprar también esta resistencias (en este caso es del 60% de los costes ~ kit).
por desgracia envío de los 200 euros de resistencia no se puede hacer por correo ordinario ... por lo que puede resultar no tan conveniente.

al igual que en el otro hilo, si le gustaría tener un poco de PCB sólo tiene que añadir su nombre a la lista.

Mimmus78: 3pcs
Nuno_pt: 3pcs
msliva: 3pcs
pelule: 3pcs
klaus11: 2pcs
================= =================
restante Juntas: 6pcs

[VK5RC]

I would like 3 if available still
Mimmus78: 3pcs
Nuno_pt: 3pcs
msliva: 3pcs
pelule: 3pcs
klaus11: 2pc
VK5RC ;3pcs

[t2kv]

And I'd like two if they're still available, please:

Mimmus78: 3pcs
Nuno_pt: 3pcs
msliva: 3pcs
pelule: 3pcs
klaus11: 2pc
VK5RC ;3pcs
t2kv: 2pc

[HighVoltage]

If there are any left, I take 2 pieces, thanks


Mimmus78: 3pcs
Nuno_pt: 3pcs
msliva: 3pcs
pelule: 3pcs
klaus11: 2pc
VK5RC ;3pcs
t2kv: 2pc
HighVoltage: 2pcs

[mimmus78]

High voltage we are one out ... if they send me some more board no problem.

[David Hess]

It looks like a classic case of popcorn noise and not pink noise.  The level is awfully low and I cannot say that I have seen popcorn noise in a zener diode but the base-emitter transistor junction in series with the LTZ1000A zener could be responsible and combined with the low impedance at the collector would explain the low level.

Screening for popcorn noise is very difficult because of its often intermittent nature.  Long test times, hours to days, make it economically impractical for most customers.  I thought I remembered a note from LT about screening LTZ1000s for something but I looked and did not find it.

[Cerebus]

It looks like a classic case of popcorn noise and not pink noise.

Agreed that it looks like popcorn noise, but I think the explanation is wandering between zener and avalanche modes. Intrinsic to these low TC zeners is the balancing of the opposite tempcos of zener and avalanche effects. I think that's already been suggested hereabouts as a cause but I'm not sure. Popcorn noise is classically associated with relatively 'dirty' semiconductor processes which is why it's generally much less of a problem than it was, say, thirty years ago. We're dealing with a buried zener here, and the burying is done precisely to keep the junction away from the surface where impurities may lurk of the sort that cause popcorn noise.

[David Hess]

It looks like a classic case of popcorn noise and not pink noise.

Agreed that it looks like popcorn noise, but I think the explanation is wandering between zener and avalanche modes. Intrinsic to these low TC zeners is the balancing of the opposite tempcos of zener and avalanche effects. I think that's already been suggested hereabouts as a cause but I'm not sure. Popcorn noise is classically associated with relatively 'dirty' semiconductor processes which is why it's generally much less of a problem than it was, say, thirty years ago. We're dealing with a buried zener here, and the burying is done precisely to keep the junction away from the surface where impurities may lurk of the sort that cause popcorn noise.

The buried zener in this case is temperature compensated by the roughly -2mV/C Vbe of a series transistor so I am suggesting that the popcorn noise is actually coming from the transistor and not the buried zener reference.

[Kleinstein]

The popcorn noise could come from the zener or the transistor - hard to tell from the data.

One could modify the circuit a little and compare the jumps. If the noise is sensitive to the zener current, this would indicate the zener as a source, if it changes with the transistor current it is likely coming from the transistor. The nice thing about the LTZ1000 circuit is than one change the currents relatively independent.

Even with that level of noise the reference might still be useful - maybe a a "two value" reference. The problem would only be if the popcorn noise gets slower and one gets only one level over a longer time - you than never know if it is the upper or lower level. So the preferred way would be to reduce the size of the jumps, not making the jumps less frequent - this could make thinks worse, even if you don't see the jumps anymore.

I don't see much sense in cooling and that way possibly destroying the chip. One might even think about using it for RMS conversion instead.

[TiN]

maybe a a "two value" reference

That's assuming both levels not change over time.

I don't see much sense in cooling and that way possibly destroying the chip

I don't see either, but desired to close this topic once and for all. Maybe we could see something in doing this (not necessarily at 77K temperature). Having 24/7 cooled to -30°C unit with TEC or phase-change loop (modified window AC) isn't that difficult, if there is point in it. I've ordered copper block at CNC shop to enclose module in hermetic box to place in dewar. This way there will be less rapid shock from contact with cooling liquid.

[Cerebus]

The buried zener in this case is temperature compensated by the roughly -2mV/C Vbe of a series transistor so I am suggesting that the popcorn noise is actually coming from the transistor and not the buried zener reference.

Yes, the LTZ is using V_BE compensation (and a lot of heat) rather than relying on zener versus avalanche tempco, but the zener versus avalanche thing is still going on because that's determined by the diodes 'set point*' voltage. I'm just influenced by the fact that there's some identifiable physics that we know goes on around this point. Anyway, the only way to be sure is to try the zener current versus transistor current experiment that's been suggested.

*Awkward terminology here because you'd normally say "zener voltage" or "avalanche voltage" but both are in play.

[David Hess]

The buried zener in this case is temperature compensated by the roughly -2mV/C Vbe of a series transistor so I am suggesting that the popcorn noise is actually coming from the transistor and not the buried zener reference.

Yes, the LTZ is using V_BE compensation (and a lot of heat) rather than relying on zener versus avalanche tempco, but the zener versus avalanche thing is still going on because that's determined by the diodes 'set point*' voltage. I'm just influenced by the fact that there's some identifiable physics that we know goes on around this point. Anyway, the only way to be sure is to try the zener current versus transistor current experiment that's been suggested.

*Awkward terminology here because you'd normally say "zener voltage" or "avalanche voltage" but both are in play.

As you pointed out, the zener is buried to prevent exactly this sort of problem.  I did a search but found nothing about popcorn noise in zener diodes which may just mean that it is such a minor problem that it takes an LTZ1000 before it becomes apparent since zeners are so noisy anyway.  I thought I remembered a comment from LT about special screening options for the LTZ1000 which included low frequency noise but was not able to find it.  In transistors, it is usually more apparent because of transistor gain but in this case, the low impedance collector load limits gain.

Since access is provided to the node between the buried zener and transistor base and actually every terminal of the transistor, testing to find the source is straightforward if not easy.

Tracking down popcorn noise was my motivation for expanding my collection of Tektronix oscilloscopes to include the 7A13 (1mV/div), 7A22 (10uV/div), and AM502 (10uV/div) differential vertical amplifiers.  It took me weeks to track down the example shown below because of its low level and especially intermittent nature where it was clean for hours at a time.

[Andreas]

I did a search but found nothing about popcorn noise in zener diodes which may just mean that it is such a minor problem that it takes an LTZ1000 before it becomes apparent since zeners are so noisy anyway.

Hello,

according to Warren they formerly selected 1N82x Zeners according to low frequency (= popcorn?) noise.

https://www.febo.com/pipermail/volt-nuts/2013-January/002362.html

see also attachement with a chart recorder.

http://www.febo.com/pipermail/volt-nuts/attachments/20130126/1fdc1ce7/attachment-0001.jpg

On another source I have read that ageing of zeners is correlated with low freqency (popcorn?) noise.
(which makes sense if you equal "popcorn noise" with "impure silicon").

With best regards

Andreas

[fanOfeeDIY]

Hi mimmus78,

Thank you, if there is still left.

Mimmus78: 3pcs
Nuno_pt: 3pcs
msliva: 3pcs
pelule: 3pcs
klaus11: 2pc
VK5RC ;3pcs
t2kv: 2pc
HighVoltage: 2pcs
fanOfeeDIY: 2pcs

[David Hess]

I did a search but found nothing about popcorn noise in zener diodes which may just mean that it is such a minor problem that it takes an LTZ1000 before it becomes apparent since zeners are so noisy anyway.

according to Warren they formerly selected 1N82x Zeners according to low frequency (= popcorn?) noise.

https://www.febo.com/pipermail/volt-nuts/2013-January/002362.html

see also attachement with a chart recorder.

http://www.febo.com/pipermail/volt-nuts/attachments/20130126/1fdc1ce7/attachment-0001.jpg

On another source I have read that ageing of zeners is correlated with low freqency (popcorn?) noise.
(which makes sense if you equal "popcorn noise" with "impure silicon").

Unfortunately I do not think the post really tells us anything.  Temperature compensated zener diodes like the 1N821 to 1N829 series include a forward biased PN junction in series just like the LTZ1000A and for the same reason so the popcorn noise still could be coming from either.

Popcorn noise is low frequency but the post confuses it with 1/f noise as well and the graph in the post shows 0.033ppm versus the 0.5ppm that TiN recorded.  Does that indicate two different sources or could it just be that the transistor in the LTZ1000A is amplifying it?  I wonder what the difference in time scale means if anything.

Isn't the connection between zener aging and popcorn noise simply because it takes time to detect?