Author Topic: VoltCard design with LM399: does this look sane?  (Read 24810 times)

0 Members and 1 Guest are viewing this topic.

Offline technixTopic starter

  • Super Contributor
  • ***
  • Posts: 3508
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: VoltCard design with LM399: does this look sane?
« Reply #25 on: May 08, 2015, 12:18:48 pm »
If Iq just has to be 1mA you can set the LM317 current to 5mA using the recommended 240 ohm resistor and put a 1k8 resistor (to shunt away 4mA) in parallel with the LM399.

As for choosing the resistors, etc, you'll just have to draw out your circuit and sit down with pen, paper and calculator and compute/determine the significance of the tolerances and drifts. For the most part it just boils down to ohms law and calculations of percentages along the lines of my prior posts. I have to call it quits for this evening.

I understand that but then the board size comes into play. Notice that the board is VERY small yet?
 

Offline c4757p

  • Super Contributor
  • ***
  • Posts: 7799
  • Country: us
  • adieu
Re: VoltCard design with LM399: does this look sane?
« Reply #26 on: May 08, 2015, 12:24:01 pm »
Don't forget that despite the datasheet recommending 240 ohms, the minimum current is actually specified as 10mA, which requires a 120 ohm resistor :-\
No longer active here - try the IRC channel if you just can't be without me :)
 

Offline technixTopic starter

  • Super Contributor
  • ***
  • Posts: 3508
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: VoltCard design with LM399: does this look sane?
« Reply #27 on: May 08, 2015, 01:18:58 pm »
Don't forget that despite the datasheet recommending 240 ohms, the minimum current is actually specified as 10mA, which requires a 120 ohm resistor :-\

Nope I changed the const current source to a TL431 and a PNP current mirror. LM317 discussion no longer valid here.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14829
  • Country: de
Re: VoltCard design with LM399: does this look sane?
« Reply #28 on: May 08, 2015, 07:08:34 pm »
The current mirror from discrete transistors is not very accurate and stable. Also the circuit uses something like 2 mA of extra supply current. A minimal circuit would be a JFET with a suitable source resistor.

The simple OP-amp to generate something like a buffered 10-12 V is likely the better solution, if there is sufficient supply voltage available. As a first guess even a LM358 might do. Usually an OP07 does not hurt.

The circuit shown above may need some extra circuit for startup: A diode at die OPs output, and a resistor to 12-15 V would do it.

The circuit should still use a separate GND line for the reference and supply / heater.
 

Offline technixTopic starter

  • Super Contributor
  • ***
  • Posts: 3508
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: VoltCard design with LM399: does this look sane?
« Reply #29 on: May 08, 2015, 07:22:37 pm »
The current mirror from discrete transistors is not very accurate and stable. Also the circuit uses something like 2 mA of extra supply current. A minimal circuit would be a JFET with a suitable source resistor.

The simple OP-amp to generate something like a buffered 10-12 V is likely the better solution, if there is sufficient supply voltage available. As a first guess even a LM358 might do. Usually an OP07 does not hurt.

The circuit shown above may need some extra circuit for startup: A diode at die OPs output, and a resistor to 12-15 V would do it.

The circuit should still use a separate GND line for the reference and supply / heater.

That current mirror is not two discrete parts, but a matched PNP pair internally connected into a current mirror (the 4-pin SOT23 on the PCB) so I think that is not really a problem any more.

I am looking at generating 10V from the reference, but I need to figure a way to do so with a trim pot but not sacrificing stability.
 

Offline technixTopic starter

  • Super Contributor
  • ***
  • Posts: 3508
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: VoltCard design with LM399: does this look sane?
« Reply #30 on: May 08, 2015, 07:42:47 pm »
Here is another revision that probably will work better.

The trim pot works, but I think the temp co of the pot does not really matter here because the entire pot drifts mostly uniformly and it is all about the ratio on this pot - the temp co cancels itself out. Maybe the wiper resistance will bite but the offset current of the op amp is low enough to prevent it from doing anything too significant.

I don't really have split ground in this project including the heater, so it is all about the layout when talking abou the heater. The PCB have GND on is front, and +12V on its back , so I think the heater will not contribute too much to the noise. Also note the extra decoupling at the op amp?
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14829
  • Country: de
Re: VoltCard design with LM399: does this look sane?
« Reply #31 on: May 08, 2015, 07:55:36 pm »
The separate ground for the heater is nor because of noise, but because of the DC feed through of the heater current, because of wire and contact resistances. At low temperature the heater may take up more than 50 mA, so that 10 mOhms of resistance at the ground plane would cause an avoidable error of something like 500 µV. In this case it would be likely better to turn of the heater all together.

So using a separate return pin is not an option. There is essentially no way around having an extra "ground" pin for the ref. output.
 

Offline technixTopic starter

  • Super Contributor
  • ***
  • Posts: 3508
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: VoltCard design with LM399: does this look sane?
« Reply #32 on: May 08, 2015, 08:05:58 pm »
The separate ground for the heater is nor because of noise, but because of the DC feed through of the heater current, because of wire and contact resistances. At low temperature the heater may take up more than 50 mA, so that 10 mOhms of resistance at the ground plane would cause an avoidable error of something like 500 µV. In this case it would be likely better to turn of the heater all together.

So using a separate return pin is not an option. There is essentially no way around having an extra "ground" pin for the ref. output.

Come to think it again, that just doesn't matter at all. If this inrush exists this device have not warmed up yet! The part I bought have a built in heat shield.

I just need to whack a line like "let the VoltCard warm up for 30 minutes after turning on before start using it" or so on to the manual of the thing.
« Last Edit: May 08, 2015, 08:31:20 pm by technix »
 

Offline GK

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: VoltCard design with LM399: does this look sane?
« Reply #33 on: May 08, 2015, 11:44:15 pm »
Don't forget that despite the datasheet recommending 240 ohms, the minimum current is actually specified as 10mA, which requires a 120 ohm resistor :-\

I read the ST datasheet lest night and that specifies 3.5mA typical, 5mA worse case:
http://www.st.com/st-web-ui/static/active/en/resource/technical/document/datasheet/CD00000455.pdf

However the spec is all over the place:

And Fairchild specify 3.5mA typical, 12mA worse case
https://www.fairchildsemi.com/datasheets/LM/LM317.pdf

TI specify 3.5mA typical, 10mA worse case
http://www.ti.com/lit/ds/symlink/lm317.pdf

but only 5mA worse case for the LM317L
http://www.ti.com/lit/ds/symlink/lm317l-n.pdf

Though Onsemi specify 10mA for the LM317L
http://www.onsemi.com/pub_link/Collateral/LM317L-D.PDF




Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline GK

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: VoltCard design with LM399: does this look sane?
« Reply #34 on: May 08, 2015, 11:52:56 pm »
Here is another revision that probably will work better.

The trim pot works, but I think the temp co of the pot does not really matter here because the entire pot drifts mostly uniformly and it is all about the ratio on this pot - the temp co cancels itself out. Maybe the wiper resistance will bite but the offset current of the op amp is low enough to prevent it from doing anything too significant.

I don't really have split ground in this project including the heater, so it is all about the layout when talking abou the heater. The PCB have GND on is front, and +12V on its back , so I think the heater will not contribute too much to the noise. Also note the extra decoupling at the op amp?


That circuit isn't going to work well at all. Your trim-pot is going to adjust the op-amp gain from AVol to unity. How are you going to accurately trim the output voltage with that? Ideally, you want an adjustment range just sufficient to cover the worse case error due to component tolerances. Study the datasheet circuit I posted in reply#8 to see how it is done properly. Also, your supply V is 12V and you have a 1k resistor in series with the LM399 zener. How is that going to give you your desired 1mA Iq?
 

« Last Edit: May 08, 2015, 11:57:24 pm by GK »
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline paulie

  • Frequent Contributor
  • **
  • !
  • Posts: 849
  • Country: us
Re: VoltCard design with LM399: does this look sane?
« Reply #35 on: May 08, 2015, 11:53:38 pm »
If this inrush exists this device have not warmed up yet! The part I bought have a built in heat shield.

I just need to whack a line like "let the VoltCard warm up for 30 minutes after turning on before start using it" or so on to the manual of the thing.

Regarding LM399 inrush 5 seconds might be more reasonable. The current on mine drops and stabilizes after about 3 seconds.
 

Offline technixTopic starter

  • Super Contributor
  • ***
  • Posts: 3508
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: VoltCard design with LM399: does this look sane?
« Reply #36 on: May 09, 2015, 01:51:12 am »
Here is another revision that probably will work better.

The trim pot works, but I think the temp co of the pot does not really matter here because the entire pot drifts mostly uniformly and it is all about the ratio on this pot - the temp co cancels itself out. Maybe the wiper resistance will bite but the offset current of the op amp is low enough to prevent it from doing anything too significant.

I don't really have split ground in this project including the heater, so it is all about the layout when talking abou the heater. The PCB have GND on is front, and +12V on its back , so I think the heater will not contribute too much to the noise. Also note the extra decoupling at the op amp?


That circuit isn't going to work well at all. Your trim-pot is going to adjust the op-amp gain from AVol to unity. How are you going to accurately trim the output voltage with that? Ideally, you want an adjustment range just sufficient to cover the worse case error due to component tolerances. Study the datasheet circuit I posted in reply#8 to see how it is done properly. Also, your supply V is 12V and you have a 1k resistor in series with the LM399 zener. How is that going to give you your desired 1mA Iq?

I cannot find SMT resistors with low enough tempco with a reasonable price and size. That is why I'd rather not use that circuit, due to form factor limitations. This circuit may be a bit tricky to adjust, but it is stable and cheaper to build, and it does not occupy too much board space.

About that 1k resistor, noticed the current mirror circuit?
 

Offline GK

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: VoltCard design with LM399: does this look sane?
« Reply #37 on: May 09, 2015, 02:07:01 am »
Here is another revision that probably will work better.

The trim pot works, but I think the temp co of the pot does not really matter here because the entire pot drifts mostly uniformly and it is all about the ratio on this pot - the temp co cancels itself out. Maybe the wiper resistance will bite but the offset current of the op amp is low enough to prevent it from doing anything too significant.

I don't really have split ground in this project including the heater, so it is all about the layout when talking abou the heater. The PCB have GND on is front, and +12V on its back , so I think the heater will not contribute too much to the noise. Also note the extra decoupling at the op amp?


That circuit isn't going to work well at all. Your trim-pot is going to adjust the op-amp gain from AVol to unity. How are you going to accurately trim the output voltage with that? Ideally, you want an adjustment range just sufficient to cover the worse case error due to component tolerances. Study the datasheet circuit I posted in reply#8 to see how it is done properly. Also, your supply V is 12V and you have a 1k resistor in series with the LM399 zener. How is that going to give you your desired 1mA Iq?

I cannot find SMT resistors with low enough tempco with a reasonable price and size. That is why I'd rather not use that circuit, due to form factor limitations. This circuit may be a bit tricky to adjust, but it is stable and cheaper to build, and it does not occupy too much board space.

About that 1k resistor, noticed the current mirror circuit?


I don't know why you keep mentioning form factor limitations given the amount of unutilized space on your PCB. Your circuit will be more than a bit tricky to adjust with any degree of accuracy. Yes, I noticed the current mirror circuit; early morning brain fart on the 1k.

« Last Edit: May 09, 2015, 02:09:02 am by GK »
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline technixTopic starter

  • Super Contributor
  • ***
  • Posts: 3508
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: VoltCard design with LM399: does this look sane?
« Reply #38 on: May 09, 2015, 02:13:47 am »
Here is another revision that probably will work better.

The trim pot works, but I think the temp co of the pot does not really matter here because the entire pot drifts mostly uniformly and it is all about the ratio on this pot - the temp co cancels itself out. Maybe the wiper resistance will bite but the offset current of the op amp is low enough to prevent it from doing anything too significant.

I don't really have split ground in this project including the heater, so it is all about the layout when talking abou the heater. The PCB have GND on is front, and +12V on its back , so I think the heater will not contribute too much to the noise. Also note the extra decoupling at the op amp?


That circuit isn't going to work well at all. Your trim-pot is going to adjust the op-amp gain from AVol to unity. How are you going to accurately trim the output voltage with that? Ideally, you want an adjustment range just sufficient to cover the worse case error due to component tolerances. Study the datasheet circuit I posted in reply#8 to see how it is done properly. Also, your supply V is 12V and you have a 1k resistor in series with the LM399 zener. How is that going to give you your desired 1mA Iq?

I cannot find SMT resistors with low enough tempco with a reasonable price and size. That is why I'd rather not use that circuit, due to form factor limitations. This circuit may be a bit tricky to adjust, but it is stable and cheaper to build, and it does not occupy too much board space.

About that 1k resistor, noticed the current mirror circuit?


I don't know why you keep mentioning form factor limitations given the amount of unutilized space on your PCB. Your circuit will be more than a bit tricky to

I think I have mentioned that SMT resistors with the required less than 2ppm temp co are difficult to find and expensive like hell. Using through hole parts is not exactly a solution as the remaining space on the board won't really fit any through hole resistors at all. The Vishay SMR1 resistors are even more expensive than the second hand LM399 reference chip. I want to minimize precision parts used on this board.
« Last Edit: May 09, 2015, 02:16:44 am by technix »
 

Offline GK

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: VoltCard design with LM399: does this look sane?
« Reply #39 on: May 09, 2015, 02:32:06 am »
If you rearranged the layout you could fit half a dozen through-hole resistors on that board without issue.  :-// To what degree of accuracy do you think you are going to be able to trim the 10V to? For a large proportion of its range that trim-pot will have the op-amp output slammed up against the positive rail. As you adjust through the targeted 10V you'll be able to see the step resolution of the wire-wound trim-pot.
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline GK

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: VoltCard design with LM399: does this look sane?
« Reply #40 on: May 09, 2015, 02:55:57 am »
A more practical circuit for trimming the output V would look like this. The trim-pot is 10k and is in parallel with 100R, which largely mitigates the trim-pots large temperature co-efficient and the adjustment range can be reduced to the required +/- a few %.

Also, using exclusively, say, just for example, either 25ppm or 50ppm resistors in this circuit will not respectively automatically return an output drift of 50ppm or 100ppm. The actual drift contribution from the feedback network will be somewhat lower due to the fact that the resistors will drift by the same proportion if in the same thermal environment.   
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Offline technixTopic starter

  • Super Contributor
  • ***
  • Posts: 3508
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: VoltCard design with LM399: does this look sane?
« Reply #41 on: May 09, 2015, 06:20:48 am »
A more practical circuit for trimming the output V would look like this. The trim-pot is 10k and is in parallel with 100R, which largely mitigates the trim-pots large temperature co-efficient and the adjustment range can be reduced to the required +/- a few %.

Also, using exclusively, say, just for example, either 25ppm or 50ppm resistors in this circuit will not respectively automatically return an output drift of 50ppm or 100ppm. The actual drift contribution from the feedback network will be somewhat lower due to the fact that the resistors will drift by the same proportion if in the same thermal environment.   

So using this design if I got all 25ppm SMD precision resistors from the same series the tempco will cancel out right?
 

Offline kjs

  • Regular Contributor
  • *
  • Posts: 97
  • Country: de
Re: VoltCard design with LM399: does this look sane?
« Reply #42 on: May 09, 2015, 06:31:11 am »
So using this design if I got all 25ppm SMD precision resistors from the same series the tempco will cancel out right?

As usual: it depends... Resistors are not like semiconductors where there is some uniformity on the wafer. They will only be within the +/- limit but the curves can be quite different. We found that it would be a crap-shot to rely on that. One batch it will work and then you have several ones where it doesn't. If it would be that easy there wouldn't be a need for multiple resistors on one substrate to get close matching of tk.
 

Offline technixTopic starter

  • Super Contributor
  • ***
  • Posts: 3508
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: VoltCard design with LM399: does this look sane?
« Reply #43 on: May 09, 2015, 06:56:54 am »
So using this design if I got all 25ppm SMD precision resistors from the same series the tempco will cancel out right?

As usual: it depends... Resistors are not like semiconductors where there is some uniformity on the wafer. They will only be within the +/- limit but the curves can be quite different. We found that it would be a crap-shot to rely on that. One batch it will work and then you have several ones where it doesn't. If it would be that easy there wouldn't be a need for multiple resistors on one substrate to get close matching of tk.

So how should I get those matched resistors? One of those precision low tempco resistor network? Those are expensive too.
 

Offline Vgkid

  • Super Contributor
  • ***
  • Posts: 2727
  • Country: us
Re: VoltCard design with LM399: does this look sane?
« Reply #44 on: May 09, 2015, 07:14:17 am »
you can easily hand match resistors, use a wheatstone bridge circuit.
If you own any North Hills Electronics gear, message me. L&N Fan
 

Offline kjs

  • Regular Contributor
  • *
  • Posts: 97
  • Country: de
Re: VoltCard design with LM399: does this look sane?
« Reply #45 on: May 09, 2015, 07:20:57 am »
you can easily hand match resistors, use a wheatstone bridge circuit.
For one-offs that's what I do but technix needs a lot of matching ones and then running the resistors over temperature and picking matching ones is rather time consuming.
 

Offline technixTopic starter

  • Super Contributor
  • ***
  • Posts: 3508
  • Country: cn
  • From Shanghai With Love
    • My Untitled Blog
Re: VoltCard design with LM399: does this look sane?
« Reply #46 on: May 09, 2015, 07:42:50 am »
you can easily hand match resistors, use a wheatstone bridge circuit.

I need to match not only the value but also relative temperature coefficient.
 

Offline GK

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: VoltCard design with LM399: does this look sane?
« Reply #47 on: May 09, 2015, 08:34:54 am »
I've had a great deal of success (both for temp co. and value matching) with through-hole precision resistors adjacent each other on the same bandolier. Don't know how smd chip resistors from the same package fare in comparison.

At the end of the day you simply aren't going to make a precision-trimmed 10V reference with true single-digit-ppm drift based on the LM399 without either some degree of component sorting or forking over the bucks for extremely stable resistors.

The LM399 is old hat anyway and there are numerous other parts available these days. Just look at the ADR01 for example:

http://www.farnell.com/datasheets/1536598.pdf

It's about 1/3rd the price of an LM399, has 3ppm typical drift for the SOIC package (1ppm for the more expensive B grade if you can source it) and the output voltage is 10V with 5mV initial accuracy - and the output is easily trimmed.
« Last Edit: May 09, 2015, 08:37:54 am by GK »
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14829
  • Country: de
Re: VoltCard design with LM399: does this look sane?
« Reply #48 on: May 09, 2015, 08:42:24 am »
It is not only the inrush current that matters for the LM399. Even under normal operation the current depends on the external temperature, the wire resistance usually has a TK of some 3000 ppm/K and the contact resistance is not that stable and predictable. So to get a reliable reference there is not way around having a separate ground connection for the voltage signal and the power. You might get away with the 1 mA current for the Ref. over the voltage contact, but even this is better avoided.

For stability of the resistors it's also important to look at humidity - some resistors also change there values with humidity. Also SMD resistors have the additional problem, that expansion (thermal and humidity) of the substrate causes stress in the SMD parts and thus can influence values.

The ADR... Refs. in SMD case are not a real alternative they have significant more long term drift and hysteresis. Also humidity is likely a problem.
 

Offline GK

  • Super Contributor
  • ***
  • Posts: 2607
  • Country: au
Re: VoltCard design with LM399: does this look sane?
« Reply #49 on: May 09, 2015, 08:50:27 am »
It is not only the inrush current that matters for the LM399. Even under normal operation the current depends on the external temperature, the wire resistance usually has a TK of some 3000 ppm/K and the contact resistance is not that stable and predictable. So to get a reliable reference there is not way around having a separate ground connection for the voltage signal and the power. You might get away with the 1 mA current for the Ref. over the voltage contact, but even this is better avoided.


The negative heater terminal of the LM399 can be tied negative with respect to the anode of the zener by a substantial number of volts and the heater can accommodate a supply voltage as high as 40V. So if you can furnish split (+/-) supply rails, the heater can be powered directly between them, keeping the heater current out of the ground completely.


The ADR... Refs. in SMD case are not a real alternative they have significant more long term drift and hysteresis. Also humidity is likely a problem.


The ARD01 will deliver better short and long term performance than an LM399-based reference followed up by a simple and cheap voltage-scaling amplifier.
Bzzzzt. No longer care, over this forum shit.........ZZzzzzzzzzzzzzzzz
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf