what is for "CURRENT REGULATOR DIODE"?

[001]

what is for "CURRENT REGULATOR DIODE"? 
Can I replace 1N5303 with some transistor circuit?

I see https://en.wikipedia.org/wiki/Constant-current_diode but how to calculate values? 

[rstofer]

The current limiting diode attempts to provide a constant current flow over a large voltage range.  If you need a current source, this is one way to get it.

I'm not going to do the math but it seems like you can achieve the same thing with a zener diode and transistor and maybe even just a transistor.  Or substitute an LED for the zener  diode.

https://en.wikipedia.org/wiki/Current_source#Simple_transistor_current_sources

[Zero999]

It's just a J-FET with its gate tied to the source and is selected to give 1.6mA.
https://www.mouser.co.uk/datasheet/2/68/1n5283-5314-43522.pdf

You can make your own with a J-FET. Most J-FETs have a higher current than 1.6mA, with zero gate voltage. To get such a low current, you'll have to connect the source in series with a resistor. J-FETs have a wide spread of characteristics, so the resistor value to get 1.6mA will have to be determined experimentally.

[exe]

These are not on sales anymore But discrete jfets are still possible to buy, though there are not that many around.

[Benta]

1N5303 is available from several suppliers:
Central Semiconductor, Digitron, Microsemi...
You can make current regulator diodes yourself from JFETs, but the current spread is huge (1:3 or more).
The "ready made" types are within +/-10%, much easier.

[gf]

I'm wondering what's the frequency response of the output impedance and the noise performance of these components? Looking at a couple of data sheets I could not find this information. Are they e.g. suitable for biasing RF amplifiers?

[David Hess]

I'm wondering what's the frequency response of the output impedance and the noise performance of these components? Looking at a couple of data sheets I could not find this information. Are they e.g. suitable for biasing RF amplifiers?

They have the output capacitance of a typical JFET but can always be used with a cascode transistor to raise the AC and DC output impedance.

[exe]

1N5303 is available from several suppliers:
Central Semiconductor, Digitron, Microsemi...

I'm not sure if it's easy to buy from them and it's worth it.They are manufacturers, I don't think they even deal with individual customers. Also, afaik, both known to produce expensive absolete parts that are not produced by other companies. They don't even have prices on their website.

Mouser lists 1N5303 for ~2.5euro (I guess VAT is not included) and as not normally stocked. Compare to much cheaper fets.

So, they are in production, but hard to buy anyway. I've also seen them on ebay, but price was what I'd call unreasonable.

[JacobPilsen]

What about CL20M45 ?

[exe]

What about CL20M45 ?

My best guess it's a normal voltage regulator used as a current regulator (google "lm317 current regulator"). There are several similar devices, normally called "led drivers" or something. I also have NSI45020AT1G (never actually tried it) . There are many variants: https://www.onsemi.com/PowerSolutions/parametrics/197/products .

They differ from a jfet in two ways: 1) much lower bandwidth 2) much higher compliance voltage (which is bad). Fets can go down well below 1V, while these regulators require a few volts to work.

[Benta]

1N5303 is available from several suppliers:
Central Semiconductor, Digitron, Microsemi...

I'm not sure if it's easy to buy from them and it's worth it.They are manufacturers, I don't think they even deal with individual customers.

The standard way to solve this is to got to their web site and check their sales channels for distributors. No semiconductor manufacturer I know of sell to SMBs or individuals.

[exe]

The standard way to solve this is to got to their web site and check their sales channels for distributors.

You're right. I said they are unobtanium because most big distributors I know don't have CRD in their inventory (digikey, mouser, tme, etc), or they cost a fortune. However, less popular distributors do have them. Like here: https://www.futureelectronics.com/search/?text=Microsemi&q=Microsemi:relevance%3AmanufacturerName%3AMicrosemi:category:current-limiting-diodes&selectedTab=products&selectedSubCategory=current-limiting-diodes&selectedParentCategory=diodes . The price is not very attractive to say the least. It can be anywhere between $2 and $200 per device.

I ended up with buying jfet from onsemi for ~$0.3, I do manual characterization and have a trim-pot for fine adjustment . Not sure if it's worth it. Probably, depends on required characteristics. The thermal stability and output impedance are not as great as I hoped. But they can have quite a small drop-out voltage. Anyway, I'm no expert, just played a bit on a breadboard.

[Calvin]

Hi,

But they can have quite a small drop-out voltage

Typically not, because to function as a proper CCS the Vds should be >2x Vgs.
Also, high-Idss/high-Vgs types degenerated by a larger source resistor are way better than types with just the desired Idss and Gate and Source shorted.
Useable standard types like the 4393 to 4391 are relatively cheap and easy to source active devices.
The 4391 is also fine as cascode for a 4392 or 4393, raising the output impedance very high .... drawback is the required high Vds voltage of > 5-10V.

regards
Calvin

[David Hess]

But they can have quite a small drop-out voltage

Typically not, because to function as a proper CCS the Vds should be >2x Vgs.

The most common discrete JFET based constant current diodes operate with a Vgs of 0 volts (the gate is tied to the source) so they do indeed have a very low dropout voltage which is limited by their Vds saturation voltage.  This is reflected by their specifications where low current diodes have a low dropout voltage (below 1 volt) and high current diodes have a high dropout voltage because the dropout voltage is the product of the current and channel resistance.

[Calvin]

Hi,

yes You´re right about Vgs=0 .... but that´s also the reason why those ´constant current diodes´are rather quick´ndirty ccs that by far don´t exploit t he potential of a proper degenerated JFET based ccs.

regards
Calvin

[001]

thank You!
its so exotic element

[David Hess]

yes You´re right about Vgs=0 .... but that´s also the reason why those ´constant current diodes´are rather quick´ndirty ccs that by far don´t exploit t he potential of a proper degenerated JFET based ccs.

I have used them to increase the line regulation of shunt voltage references to good effect but availability is always a problem so it is usually better to use a discrete current regulator circuit.