-
I dont want to overload the KX thread so i started a fresh thread for my problem!
Finally i found some time to assemble one of the KX-boards that were lying around for some months already.
I had a really hard time because i don´t have any experience with these small smd parts. I really should have practised with some trash parts before risking the damage of expensive components.
Anyway, i managed to solder the parts on the PCB somehow.
The two biggest mistakes were :
a) Mounting LT1013 instead of LTC2057!
( dont ask me why, i had them both lying in the component-box for KX boards and totally forgot that i bought the LT1013 for some future project. And they are ( of course ) not pin compatible)
b) dropping the assembled board on the floor while i was trying hard de desolder the LT1013´s
Actually i discovered the wrong Opamp after i tried to power up the board and Zener Vref was some millivolt.
Then i replace them with the correct LTC2057 and after powerup i could measure 7.15xxxxx Volts.
That looked promising but the output was still unstable after the 4th digit and so i did some more inspections.
I changed two of the film capacitors ( only suspected visible damage ) and reflowed some components that appeared to not have proper soldering. I also found some very tiny traces of solder shorting some tracks on the PCB and removed those.
After that the output was down to 7,09xxx Volt but stable to the last digit ( 6.5 Digit meter ).
But even after one hour of operation i could not feel any warming of the LTZ1000 casing.
I then found out ( by measuring the potential of some points ) that no real current is flowing though the heater ( only 1 mV drop across pin 1 and 2 of LTZ, heater has around 300 Ohms -> 3 microampere current through heater ? ).
Here is the circuit from AN42 with my measurements :
Sorry, unable to do inline images. Pic is attached now!!!
Yes, my knowledgement is still pretty limited. I am studying a lot of books right now to catch up.
But it seems to me that the zener part of the LTZ is working -> 7.09 across zener and base-emitter of Q1 of the LTZ.
But the temperature control part seems like having problems. there is 550 mV across - and + of the opamp. So that opamp should be saturated right?
Meaning the output of the opamp is correct?
Why is the voltage drop across collector emitter of Q2 of LTZ only 41mV?
So should i suspect the Q2 of LTZ is damaged?
Any suggestions?
They are all very welcome!!!
Thank you very much
Joerg
-
Close-up photos of the board might help. You can upload them to FTP (linked in my sub), I'll host them for you.
Check resistance across all film capacitors. They are very easy to damage by heat, maximum temperature is only 160C.
If you have one of the caps tied to opamp inputs shorted to return path, you'd see such condition. Also do not run the circuit when heater is stuck at max power, you can easily overheat and destroy the LTZ oven. That might already happen if you let it run like this more than few seconds, if you used LTZ1000ACH chip and not LTZ1000CH. -
R4 controls the regulation temperature of the chip. The voltage shown on the schematic (0.60v on the base of Q2) indicates that R4 is a 12k resistor. Can you temporarily add 1000 ohms in series with R4 to see if that raises the temperature control point so the heater starts to work?
The voltage on the collector of Q2 is absolutely correct if the ambient temperature is to high for the R4 value selected. HP used a value of 15k for R4 but then the chip runs at 95 deg C and drifts more than it needs to.
During development I run the power supply down at 10-12v to reduce the risk of overheating the expensive zener chip. -
Hello,
I would also check if R3 is open (has infinite resistance) instead of 70K.
with best regards
Andreas
-
By the way:
the first thing that I bought for precision references is a complete rework-station.
(hot air soldering + desoldering iron with integrated vacuum solder pump)
Otherwise the chance to damage something is relatively high.
with best regards
Andreas
-
Thanks to all for your suggestions.
I´m not at home right now. I checked all resitances and also the caps .
As far as i recall there was no short.
Will check it again when home, and also try some pics of the board.
The heater should be okay, i measured around 300 Ohms.
And to me it looks like stuck at minimum , not maximum heater power.
Yes i am looking for a rework station already.
Because i am sure i will do more smd projects in the future.
Thanks
Joerg -
Just a short update...
I checked everything again.
Then ( i thought ) i found a short between non-inverting input of the Op-amp and Ground.
I desoldered and soldered op-amp two times because i just couldn´t get it clean.
There is a very tiny trace between pin 3 and 4 on the PCB and with my tools it is very hard to do it.
Since i had no success with that board i decided to assemble a fresh one with fresh parts.
( i bought components for 4 boards ).
This time it was a bit easier since i already had some experience.
The outcome was exactly the same as my first board.
Then i realized that i just copy&paste the list of materials from TIN´s website.
So i bought 10K and 1K for the voltage divider that determines the set temperature for LTZ chip.
Since i had no 12K or 13K resistor i used cheap metal-film 10K + 2,2K.
The board is running as expected now. LTZ casing is about handwarm ( 33 celsius quick theromcouple measurement).
Would 13K resistor for the divider be enough?
Or should i go for 13,5 or 14?
Do you think the behaviour for 10K is correct? No heater power?
Thanks again to all.
Joerg -
I would not use a ratio any less then 12.5:1 for any normal room temp environments. I don't think you mentioned it, are you using the LTZ1000 or the LTZ1000A? The "A" should be set to run a bit warmer as the internal heating and higher thermal resistance causes the chip to run warmer giving you less headroom for external environmental conditions. 13K/1K is what I run on my three LTZ1000A's, they are good up to about 45deg C ambient with no real loss of internal temp regulation.
-
12.5K/1K works fine if you can make sure ambient temperature never goes over +35c, otherwise 13-13.5K/1K would be better.Quote
Then i realized that i just copy&paste the list of materials from TIN´s website.
Sorry about that, already fixed BOM for 13K part instead to avoid future mishaps.
So i bought 10K and 1K for the voltage divider that determines the set temperature for LTZ chip. -
...
Sorry about that, already fixed BOM for 13K part instead to avoid future mishaps.
No need to say sorry. My mistake.
Temp setting is the first thing one should think about before building this thing.
So i obviously skipped my homework there
By the way, i think there is a typo in the schematics.
Transistor PZT3904 should have emitter on PIN 3.
Regards
Joerg -
Quote
By the way, i think there is a typo in the schematics.
SOT-223 footprint on my board have 1, 2 pins, and pin 3 is tab terminal.
Transistor PZT3904 should have emitter on PIN 3. -
It is pin compatible. I was referring to PCB layout numbering, not the component. Datasheet I used did not have numbers, so you can all-together ignore what I said
.
Traditional SOT-223 transistor working here just fine.
-
I am thinking about to reuse the 10K precision resistors that i bought.
In theory i could accompany them with a 770 Ohm resistor.
That should be the same ratio as 13K/1K right?
Did i miss anything?
Regards
Joerg
-
A 770 ohm precision resistor would work.
Another option is to use a 3,333 ohm resistor in parallel with the 1k resistor. This would achieve the same results.
Good luck! -
I dont wanted to just sit and wait until the 13K resistors arrive.
So i did some data logging in the mean time to see how the circuit "behaves".
First of all some pictures of the assembled board.
You can see the "bodged together" subsitute for the missing 13K resistor.
-
This is some data logging of the Vref.
Board is supplied by 12V lead acid battery.
Time span is 30 min. Around 90 samples at 20s integration time.
-
Of course i have a lot of questions now!!! ( surprise
)
First question is how much improvement should i expect from the ordered 13K vishay Z201 resistor?
Of course the vishay will have superior TC.
But will it help me eliminate the noise?!
The package from texascomponents arrived today but unfortunately they sent me the wrong type.
Instead of the Z201 that i ordered i got TX2575.
Not sure if i should use those. And before i try i have to solve that issue of wrong shipping with the company.
-
Next question is the performance of my meter.
As i am "comparing" the KX output with the reference of my meter ( PREMA 6048 ) i wonder who is contributing what amount of noise/drift.
In some other thread i found the advise to check the noise of the meter with a 9V battery.
That doesnt work for me because all of my 9V blocks do have too much noise.
So i took 4 NIMH cells and logged their voltage over 5 hrs ( 20sec int; 900 samples )
That looked like straight line to me.
Though i am not sure if the rate of voltage drop is normal selfdischarge or is it due to some leakage at the input of the meter?
-
I "converted" that graph to have an idea of the noise.
I just subtracted each n from n+1 in the spreadsheet.
C1 = B2 - B1
That produced the following plot.
Average = 0,0000009 V
Min = 0,0000002 V
Max = 0,0000016 V
SD = 0,0000002 V
Now i am unsure if this is a valid method to measure ( or "guestimate" ) the noise of my meter in the DC voltage circuit.
Experts please !!!!
Regards
Joerg
-
I'm not familiar with your Prema meter, but noise at first glance looks okay to me.
Typical LTZ reference 7V noise is ~1.0-1.6 uV. In the end , you have noise from own meter reference, and then noise from the DUT LTZ reference. -
Of course i have a lot of questions now!!! ( surprise
)
First question is how much improvement should i expect from the ordered 13K vishay Z201 resistor?
Of course the vishay will have superior TC.
But will it help me eliminate the noise?!
The package from texascomponents arrived today but unfortunately they sent me the wrong type.
Instead of the Z201 that i ordered i got TX2575.
Not sure if i should use those. And before i try i have to solve that issue of wrong shipping with the company.
The TX2575 is superior to the Z201 in the face of changing humidity. The epoxy over-mold on the Z201 absorbs water vapor, then swells because of it, and that strains the resistor. You can see around 25ppm of change from low humidity to high humidity (which is reversible). Most of this humidity sensitivity is gone with the TX2575, because the only epoxy in it is used to glue the foil to the ceramic substrate-- which IS affected by humidity, but not by very much as it is very thin. So, I would say go ahead and use the TX2575 resistors, but you can make a deal with Texas Components to refund you any price difference (or give you store credit, at least). Handle the TX2575 with great care though-- they are easily damaged by rough handling. -
I'm not familiar with your Prema meter, but noise at first glance looks okay to me.
Typical LTZ reference 7V noise is ~1.0-1.6 uV. In the end , you have noise from own meter reference, and then noise from the DUT LTZ reference.
I agree, it does not look completely off!
In theory the noise of my meter from the battery testing seems like +/- 0,7 microvolt.
Together with the LTZ spec. noise +/- 1,5 microvolt would still be "normal".
But it seems to be possible to get that noise down.
For example: https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg268035/#msg268035
And i am not complaining about the fast spikes.
I dont like the "slow" oscillating/drifting and i wonder if this is some problem of my build.
Bad solder joint, improper shielding, etc ....
By the way...
How can you achieve the noise spec of a HP3458A in the 10 Volt Range ( 0,01 ppm RMS ) if your LTZ inside the HP3458A has a noise of 1.0 - 1.6 uV?
Or am i doing wrong calculations here?
About the TX2575 Resistors:
Texas Components offered me a very generous solution.
They will send me the correct type for free...
... and i can keep the TX2575.
That is extraordinary customer service!!!
I can highly recommend this company!!!
Joerg
-
This is some data logging of the Vref.
I think KX-board is Ok, but Prema 6048 is too noisy meter. 0.3 ppm p-p noise is rather large for 20 s integration time and 30 min. span. 2 ppm / 84 hrs drift is too large also.
Board is supplied by 12V lead acid battery.
Time span is 30 min. Around 90 samples at 20s integration time.