Hi guys, a physics question for you. I was looking for a diode with a very low dynamic impedance. This is the delta-V forward drop resulting from a delta-I forward current.
Then I wondered: Is the diode forward voltage drop versus forward current purely physics determined or are there also manufacturing factors involved? In other words, is my search futile because all diodes are the same in this respect?
Jan
Then I wondered: Is the diode forward voltage drop versus forward current purely physics determined or are there also manufacturing factors involved? In other words, is my search futile because all diodes are the same in this respect?
Jan
Larger and larger current rated silicon diodes will have lower and lower "rd" - the incremental resistance at highish current. Large industrial diodes are characterised by a threshold voltage Vt and an rd, such that the resulting diode on-voltage is Vt + rd.Id, and any datasheet will include those values, plus a variety of effective thermal resistances to use for different current waveform often found, and different cooling schemes that conduct heat from anode or cathode or both., along with transient thermal resistance curves/equations.
So this is basic silicon thickness and doping resistance and conduction area, and relates to PIV.
So this is basic silicon thickness and doping resistance and conduction area, and relates to PIV.
Last edited:
In my early career selenium rectifiers were just being replaced by junction rectifiers. The selenium were awfully lossy of course and the internal resistance was touted as a current limiter. For very high voltage and current we used mercury arc tubes.
At the same time for low power and signal applications I recall a variety of cat's whisker types ("crystal set" radio receivers etc) with very low forward voltage. Some were crystal based (galena etc) and others silicon or germanium but all had a claimed low forward drop. Maybe that was a simple comparison with selenium.
Just wondering out loud if the WW2 communications cats whisker diodes might have had the required physics.
At the same time for low power and signal applications I recall a variety of cat's whisker types ("crystal set" radio receivers etc) with very low forward voltage. Some were crystal based (galena etc) and others silicon or germanium but all had a claimed low forward drop. Maybe that was a simple comparison with selenium.
Just wondering out loud if the WW2 communications cats whisker diodes might have had the required physics.
Yes Johno I have one taken from a military communications receiver a bit bulky and old school but I will look it out and test it.
Schottky Diodes
Schottky Diodes
Last edited:
It is determined by the resistivity of the starting wafer, the diffusion profile of the diode, the base width at the junction, and the area of the final die. It is very dependent on the final application desired.
So, tremendously dependent on the manufacturing diffusion process.
Jn
(I have to say I am not a part designer).
If we are talking about rectifier diodes - manufacturing factors are involved in a small portion - such a way that designers may use thicker wires or thinner wires, large crystals or smaller crystals for the same current. So we'll have larger total resistance with the same current or lower. Maximum rated voltage as you know affect forward resistance too.
Many factors are involved other than the basic (simplified) junction equation.
As pointed out, large area (=higher Is), low PIV (=lower resistivity) plus a number of non-ideality factors play a role.
If you really want a physical diode (just two wires), opt for an epitaxial diode: fast or ultra-fast diodes are a good example. (BYW31 and many others).
If you are not married to the diode form-factor, use a diode-connected transistor: transistors made for log/antilog conversion are particularly good in this respect, but a device like the ZTX851 is probably an excellent candidate too
They are not the same. For example I found that Vishay shottky diodes have a bit lower Vf than unknown Chinese ones with the same current (it is easy to check with laboratory power supply in CC mode).
For example VS-10TQ045PBF has lower voltage drop then some cheap MBR1045.
Last edited:
I hadn't thought about that, but yes maybe a transistor as diode has lower incremental impedance.
I am not worried about the resistance as such, as long as it is constant, in other words, as long as the diode reacts as much as possible as a resistor.
This probably means you need to be well beyond the knee of the diode, in the almost-straight part.
For instance, in use as a cathode 'resistor' you can handle the smallish resistance but it should be constant to avoid distortion.
Jan
I think you will find 2 components involved (and in series)
1) will be the "pure diode" for lack of a better word.
There you will have the "junction jumping" voltage drop, which is not linear, drops 2mV per deg C increase if Silicon, has a threshold (again depending on material and probably somewhat on doping) , depends on current, increasing less as you go up in current which gives its characteristic "diode curve", and am quite certain it depends on current density, so a "large" diode (say 1N5402) will drop less at a certain current level (say 50mA) than a "small" one (think 1N4148)
2) will be the purely resistive part outside of the junction area but where the current has to go through.
Base material is a semi-conductor, so a quite poor one (compared to plain metals) and its resistivity will depend on doping and maybe further treatment.
So you WILL have "a diode" but it won´t be "perfect" (although they are d*mn good) , they will not be "all the same" and there will allways be a certain plain resistive component.
Datasheets provide the basic information but for a critical design I suggest you test them yourself and try a few until you find those who fit your purpose best.
Yes, that's a nice part Bill. But as I mentioned, I don't mind the threshold - that is why I want to use a diode in the first place. What I want is a constant incremental resistance.
That resistance is the slope of the V-I curve (duhhh ...) so for that to be constant, that slope-line must be a straight line.
All your posts have helped me to fine-tune my question: from a data sheet you can see that if you push enough current through a diode, you end up high on the forward slope, where it becomes more and more straight, thus at a constant incremental impedance. Good.
Now suppose I want to have a diode that has already a constant impedance at, say, 2 - 5mA. What would I be looking for? Is there a smarter way to look at specific parameters, or am I condemned to a random search through a forest of data sheets hoping to strike gold?
Or have all diodes at 5mA similar incremental resistance, do their curves all have the same 'bend'?
Jan
That resistance is the slope of the V-I curve (duhhh ...) so for that to be constant, that slope-line must be a straight line.
All your posts have helped me to fine-tune my question: from a data sheet you can see that if you push enough current through a diode, you end up high on the forward slope, where it becomes more and more straight, thus at a constant incremental impedance. Good.
Now suppose I want to have a diode that has already a constant impedance at, say, 2 - 5mA. What would I be looking for? Is there a smarter way to look at specific parameters, or am I condemned to a random search through a forest of data sheets hoping to strike gold?
Or have all diodes at 5mA similar incremental resistance, do their curves all have the same 'bend'?
Jan
Last edited:
At some point, all diodes become resistive when their ohmic resistance overcomes the junction equation completely.
Thus, you should look for very imperfect diodes: small, HV types, but not epitaxial. BYX10 comes to mind, but it is obsolete.
PIN diodes do that very well, but only above a certain frequency.
In the spice parameters, Rs is relevant, but spice models are notoriously incorrect
Thus, you should look for very imperfect diodes: small, HV types, but not epitaxial. BYX10 comes to mind, but it is obsolete.
PIN diodes do that very well, but only above a certain frequency.
In the spice parameters, Rs is relevant, but spice models are notoriously incorrect
I have to wonder what your application is. As you might expect, telling us this will help us find a solution for you.
If you want a "fixed resistance" that changes less than the diode dynamic resistance with current change, you can add a small series resistor which adds to the dynamic resistance and makes the change in resistance a smaller percentage.
If you can stand the 2.5V voltage drop, an LT431/TL431 adjustable regulator has very low dynamic resistance (of course it's an amplifier with feedback to keep the voltage constant), and a (offhand from memory) current range of 1ma to 100mA. You can put a 1 ohm resistor in series with that, and the total resistance should be (okay, just guessing from how the device operates) dominated by the 1 ohm resistor.
Yes. Somewere about maximum forward current (and higher) it is almost straight line.
Jan, perhaps the info you need to provide is the range of forward current over which you would like the incremental resistance to be as constant as practical- is it 1uA to 1mA, or 1A to 100A, or 1uA to 1kA .....
All diodes will exhibit threshold voltage non-linearity, which also goes to how long you want to operate at the max end of your range (ie Tj change during operation), and high current non-linearity where the conducting material effectively starts to saturate.
All diodes will exhibit threshold voltage non-linearity, which also goes to how long you want to operate at the max end of your range (ie Tj change during operation), and high current non-linearity where the conducting material effectively starts to saturate.
Last edited:
Why not use a diode-connected transistor in series with a resistor then? That must be more reproducible than a diode that has a large bulk resistance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.