Spectrophotometry very stable current light source circuit

[smilem]

The PDF for MAX662A lists:
https://www.electro-tech-online.com/custompdfs/2009/05/MAX662A.pdf

The capacitance values of the charge-pump capacitors C1 and C2 are critical. Use ceramic or tantalum capacitors in the 0.22μF to 1.0μF range. For applications requiring operation over extended and/or military temperature
ranges, use 1.0μF tantalum capacitors for C1 and C2

So the diagram should have a plus sign for the shown capacitors? I mean I need polarized capacitors?

I have chosen these for C1, C2:
KEMET|T110A105K035AT|CAPACITOR, AXIAL, CASE A, 1UF | Farnell LT

Also I found:

The values of C4 and C5 can be reduced to 2μF and 1μF, respectively, when using ceramic capacitors. If using aluminum electrolytics, choose capacitance values of 10μF or larger for C4 and C5.

See figure 3b in the PDF it has the plus signs back? Also they use 22uf caps there.

So I'm confused, can anyone list best correct capacitors for my project from these 2 sites:
www.farnell.com www.elfalektronika.lt (choose language english).

 

[ccurtis]

The capacitance values of the charge-pump capacitors C1 and C2 are critical. Use ceramic or tantalum capacitors in the 0.22μF to 1.0μF range. For applications requiring operation over extended and/or military temperature
ranges, use 1.0μF tantalum capacitors for C1 and C2

So the diagram should have a plus sign for the shown capacitors? I mean I need polarized capacitors?

C1 and C2 must be in the range of .22uF to 1uF as the datasheet indicates. That's a large range, making precision capacitors (±10% tolerance is fine) unnecessary. The datasheet gives the connection polarity for capacitors that are polarized; Cx+ for positive and Cx- for negative. Generally, capacitors lower than 1uF are not polarized, so then, the connection polarity is not relevent for those.

I have chosen these for C1, C2:
KEMET|T110A105K035AT|CAPACITOR, AXIAL, CASE A, 1UF | Farnell LT

Those are fine for C1 and C2.

Also I found:

The values of C4 and C5 can be reduced to 2μF and 1μF, respectively, when using ceramic capacitors. If using aluminum electrolytics, choose capacitance values of 10μF or larger for C4 and C5.

See figure 3b in the PDF it has the plus signs back? Also they use 22uf caps there.

So I'm confused, can anyone list best correct capacitors for my project from these 2 sites:

For C4 and C5, use the same capacitor you choose for C1 and C2, but increase the value to between 4.7uF and 10uF, inclusive.

In figure 3b, the capacitors are nonpolarized, 0.22uF, so they can't be connected backwards.

 

[smilem]

So like you say for C1 and C2 I'll use tantalum

T110A105K035AT 1uF
KEMET|T110A105K035AT|CAPACITOR, AXIAL, CASE A, 1UF | Farnell LT

I can't get this brand in other size

The caps in figure 3b are polarized see picture. So to clarify things up. I need 22uf or like you said 4.7uf and 10uf?

Tantalum or electrolytics is better for C4 C5?

I'm puzzled between:
Tantalum temperature stable:
http://www.elfaelektronika.lt/artnr/-81-00-12/tantalum-capacitors-axial-t-110

Tantalum Glass:
http://www.elfaelektronika.lt/artnr...ic-capacitors-for-advanced-requirements-150d-

Also is it better to use WIMA 0.1UF 63V MKS2 RM5mm Metallized Polyester (PET)
or
http://lt.farnell.com/vishay-bc-components/225232614104/capacitor-0-1uf-100v/dp/1141783

 

[ccurtis]

I need 22uf or like you said 4.7uf and 10uf? Tantalum or electrolytic is better for C4 C5?

If tantalum, use 4.7uF-10uF. If electrolytic type, use something greater than 10uF. Tantalum capacitors have lower leakage current and lower ESR than electrolytic. You can choose larger capacitance values for C4/C5 to reduce ripple even more, but you get diminishing returns at needless loss of space and increased cost. The polarity of the capacitors in Figure 3b is still correct (sorry, I was looking at 3a).

I do not know the effects of a small amounts of low frequency ripple in the LED current in your application. If ripple is a concern then you want to use a higher capacitance value for C5, rather than a lower value. You can increase the value to a hundred uF, or more, and the circuit will work.

Also is it better to use WIMA 0.1UF 63V MKS2 RM5mm Metallized Polyester (PET) or http://lt.farnell.com/vishay-bc-com...pin, but the difference will be unnoticeable.

 

[smilem]

Thanks for help.

 

[smilem]

Hi, I got all parts and have soldered them successfully, however I get 15mA instead of required 8mA or less, my LED seems to work at 6mA too

It seems the 330ohm resistor is wrong?

I tried to add 2 resistors each 2.2K ohm in parallel to make 1K ohm.

I connected it to the power output (to protect my LED) and got 6mA but I have no idea how calculate required resistor instead of 330ohm one.

 

[Sceadwian]

If you have 6mA as it stands you meet your 8ma or less requirement, what's the problem?

 

[ccurtis]

My appologies. I used the 15 mA nominal value I read in the data sheet for the LED to select the 330 ohm resistor. Calculating the resistor you need is easy -- R(ohms)=5000÷ILED(mA), or for 8mA, R=625 ohms.

 

[smilem]

If you have 6mA as it stands you meet your 8ma or less requirement, what's the problem?

6mA with aditional 1K ohm resistor on power output.

 

[smilem]

My appologies. I used the 15 mA nominal value I read in the data sheet for the LED to select the 330 ohm resistor. Calculating the resistor you need is easy -- R(ohms)=5000÷ILED(mA), or for 8mA, R=625 ohms.

Thanks, so I should use Res MF 620ohm 0,1% 15ppm (282-5) right?

 

[Hero999]

You could use that but 0.1% is overkill, for a simple current limiting resistor 5% carbon film will do.

 

[smilem]

You could use that but 0.1% is overkill, for a simple current limiting resistor 5% carbon film will do.

I know 5% is good enough but they must be temperature stable and 5% resistors a not as far as I know.

 

[Hero999]

The forward voltage of the LED will vary more over temperature so will affect the current more than the resistor.

 

[ccurtis]

Smilem, I thought you wanted something less than 8mA, so replace the 330 ohm resistor with a different value for the current you want through the LED based on the relationship I posted. I offered 8 mA (625 ohms) only as an example. You only need a precision resistor if you need a precise value of current through the LED. Since stability with time and temp is your main requirement the precision of the ohmic value of the resistor does not affect stability of the current through the LED as long as the resistor is stable over time and temperature. The stability of the current through the LED is directly related to the stability of the resistor, so you do want a resistor with low PMM change in value over time and temperature.

 

[Hero999]

You've missed the point the LED has a far greater tolerance than even a 5% carbon film resistor so using a precision resistor is a waste of money.

 

[ccurtis]

You've missed the point the LED has a far greater tolerance than even a 5% carbon film resistor so using a precision resistor is a waste of money.

The LED is outside of the point where the current through the resistor, and thus the series LED, is sensed by the MAX device. Variance in the current through the LED due to variance in the forward voltage drop of the LED is corrected by sensing the voltage change across the resistor.

 

[smilem]

You've missed the point the LED has a far greater tolerance than even a 5% carbon film resistor so using a precision resistor is a waste of money.

The LED is outside of the point where the current through the resistor, and thus the series LED, is sensed by the MAX device. Variance in the current through the LED due to variance in the forward voltage drop of the LED is corrected by sensing the voltage change across the resistor.

Can you help me understand and select correct resistor tolerance in % (percent) and PPM value for my project?

Since the MAX chip is so precise it would be a waste to have 15ppm resistor wouldn’t it?
even if it's 0.1% tolerance right?

So what PPM resistor I need for at least two decimal place precision I’m trying to get?

For example I found: VISHAY DALE - S102JT 500R 0.01% ± 0.6ppm/°C
The bad thing is I need at least 625ohms and theese are expensive

 

[ccurtis]

Smilem, you want .06% stability for two decimal place stability: .005mA/8mA=.000625=.06%.

1. Therefore, you want 625ppm stability over the range of temperature the resistor will experience: .000625×1,000,000=625ppm.

2. Therefore, if the range of temperature the resistor will experience is 30 deg C, you want a resistor with 20.8 ppm/deg C stability: 625ppm/30=20.8 ppm.

However, the MAX device has it own stability rating over temperature which adds to the stability (or instability) of the resistor, so for adequate overall stability you subtract the stability of the MAX (5ppm/deg C) to get 15.8 ppm/deg C for the resistor: 20.8ppm-5ppm=15.8ppm

 

[Hero999]

The LED is outside of the point where the current through the resistor, and thus the series LED, is sensed by the MAX device. Variance in the current through the LED due to variance in the forward voltage drop of the LED is corrected by sensing the voltage change across the resistor.

I didn't notice this when I made my previous post but even so my comment still holds true.

Even if the current through the LED is controlled to a tolerance of 0.1%, the intensity emitted by the LED will still vary by a much larger percentage over the operating temperature range and the life of the LED.

There's no point in using a tight tolerance high stability resistor when the LED has very wide tolerance and a poor stability.

 

[ccurtis]

Even if the current through the LED is controlled to a tolerance of 0.1%, the intensity emitted by the LED will still vary by a much larger percentage over the operating temperature range and the life of the LED.

The OP wants a stable current through the LED. If that is not going to get him what he ultimately wants, such as a stable light output over time and temperature, then yes, the circuit is an exercise in futility. I only assume that the OP knows what he needs to do to get where he wants to go.