Summary of the invention
The object of the present invention is to provide a kind of low noise figure erbium-doped fiber amplifier design optimization methods, by setting
The loss of specific intergrade is counted with the variation tendency of wavelength, VGEDFA under conditions of keeping gain spectral constant or being basically unchanged,
Obtain more preferably noise figure NF (Noise Figure).The technical solution adopted by the present invention is that:
It is the formula of log unit except illustrating in the present invention;By taking two-stage amplifier as an example, for VGEDFA,
The corresponding internal gain of wavelength X are as follows:
G'(λ)=G1(λ)+G2(λ) (1)
G1(λ) is the corresponding first order amplifier gain of wavelength X, G2(λ) is the corresponding second level amplifier gain of wavelength X;
The corresponding external gain of wavelength X are as follows:
IL (λ) indicates the corresponding intergrade loss of wavelength X;
The average gain G in outside are as follows:
Wherein, G and G ' respectively indicates external average gain, internal average gain, G1Indicate that first order amplifier averagely increases
Benefit, G2Indicate that the average gain of second level amplifier, IL indicate intergrade average loss;
When multi-wavelength amplifies, GR and GT is introduced, wherein GR indicates gain fluctuation, and GT indicates gain slope;Mathematical formulae is such as
Under:
Gline(λi)=k* λi+c
GT=Gline(λn)-Gline(λ0)=k* (λn-λ0)
GR=max [G (λi)-Gline(λi)]-min[G(λi)-Gline(λi)] (4)
I=0....n
GlineIt is the linear fit function of gain and wavelength, k and c are the slope and intercept of formula (4) respectively;λiIndicate i-th
A wavelength;Start wavelength is λ0, termination wavelength is λn;GT, GR unit are dB;
For a VGEDFA, GT and G ' are linear,
GT=k2*G'+c2
=k2* (G-IL)+c2 (403)
Wherein k2 and c2 is respectively slope and intercept in formula (403), available by calibrating;K2 is negative value;
It is learnt by formula (403), if G is remained unchanged in VGEDFA, GT and IL are linear;Otherwise IL is remained unchanged,
GT and G are linear;
In VGEDFA, if wavelength XiCorresponding internal gain G ' (λi) remain unchanged, when IL changes, when:
1)ΔIL(λi)=Δ IL, i.e. IL variable quantity and wavelength are unrelated, and the corresponding intergrade loss of any one wavelength becomes
Change amount is identical, by formula (2) it is found that G (λ at this timei) it is that overall variation and wavelength are unrelated, i.e., gain spectral shape does not become
Change, GR, GT are remained unchanged;ΔIL(λi) indicate wavelength XiCorresponding intergrade Dissipation change amount, Δ IL indicate that intergrade is average
Dissipation change amount;
2)ΔIL(λi) ≠ Δ IL, i.e. IL variable quantity is related to wavelength, the corresponding intergrade Dissipation change amount of different wave length
It is inconsistent, to Δ IL (λi) and λiIt carries out linear fit and obtains linear function Δ ILline(λi), calculate the corresponding Δ of Δ IL
Tilt is indicated with formula (401);By formula (2) it is found that gain spectral shape is changed in this case: GR, GT are no longer kept
It is constant;
ΔIL(λi)=IL (λi)-ILbg(λi)
ILbgIndicate intergrade background losses, ILbg(λi) indicate wavelength XiCorresponding intergrade background losses;IL(λi) table
Show wavelength XiCorresponding intergrade loss;K3, c3 are the slope and intercept of formula (401);Due to λn≥λ0, therefore Δ Tilt and k3
Positive negativity having the same;
If Δ Tilt is the function of Δ IL, indicated with formula (405):
Δ Tilt=f (Δ ILline)=f (IL-ILbg) (405)
After introducing Δ Tilt, GT and GTbaseRelationship it is as follows:
GT=GTbase+ΔTilt
ΔGT1=GT-GTbase=Δ Tilt (402)
ILbg=Gmax-G′base (404)
GmaxIndicate maxgain value, G 'baseIndicate benchmark internal gain;GTbaseIt is defined as reference gain slope, i.e.,
ILbg、Gmax、G’baseCorresponding gain slope;
Learn that G is remained unchanged by formula (403), the variation of IL causes GT to change, if variable quantity is Δ GT2:
ΔGT2=GT-GTbase=(G'-G 'base)*k2
=[(G-IL)-(G-ILbase)]*k2 (5)
=(ILbase-IL)*k2
ILbaseIndicate G, G 'baseCorresponding benchmark intergrade average loss, i.e. ILbase=G-G 'base;Formula (402) table
Bright, the introducing of Δ Tilt will lead to GR, GT variation, and wherein the variable quantity of GT is Δ GT1;Formula (5) indicates that the variation of IL can be led
GT under identical gain is caused to change, variable quantity is Δ GT2;By GT=GTbase+ΔGT1+ΔGT2It learns, to keep GT constant,
As long as meeting formula (501);
When G is some steady state value, the influence as caused by Δ Tilt is compensated by adjusting G ' or IL, when GT variable quantity meets
When formula (501), GT is remained unchanged;
ΔGT1+ΔGT2=0 (501)
Further abbreviation obtains:
It is learnt by formula (502), the variation tendency of IL (λ) can be designed as particular value, since K2 is negative value:
1) as Δ Tilt < 0, IL is greater than ILbase;
2) as Δ Tilt > 0, IL is less than ILbase;
For two-stage amplifier, noise figure NF formula:
NF1Indicate first order amplifier noise index, NF2Indicate that second level amplifier noise index, formula are adopted in (701)
Use linear unit.
When log unit is converted to linear unit, numerical value perseverance is greater than 0;Work as IL > ILbaseWhen, NF is learnt by formula (701)
(IL)<NF(ILbase), i.e., as Δ Tilt < 0, the case where G and GT are remained unchanged, since IL value increases, NF can obtain excellent
Change;Conversely, working as IL < ILbaseWhen, NF can be deteriorated.
The present invention has the advantages that the present invention is by designing specific wavelength XiIL (λ is lost in corresponding intergradei), it is different
The corresponding intergrade Dissipation change amount Δ IL (λ of wavelengthi) different, as Δ Tilt < 0, with the reduction of gain, kept not in GT
In the case that change, GR are without obvious deterioration, noise figure NF can be optimized.
Specific embodiment
Below with reference to specific drawings and examples, the invention will be further described.
By two sections of Er-doped fibers composition VGEDFA for, as shown in Figure 1, include gain be G1 first order amplifier,
The variable intergrade for being IL is lost, the second level amplifier that gain is G2;The light channel structure of VGEDFA is in this professional domain
One of most commonly seen optical design, the equal familiar with understanding of personnel in this profession, the present invention does not introduce specially, only with schematic diagram
It indicates.
It is the formula of log unit except illustrating in the present invention;For VGEDFA,
The corresponding internal gain of wavelength X (abbreviation of internal actual gain) are as follows:
G'(λ)=G1(λ)+G2(λ) (1)
G1(λ) is the corresponding first order amplifier gain of wavelength X, G2(λ) is the corresponding second level amplifier gain of wavelength X;
The corresponding external gain of wavelength X are as follows:
G (λ)=G1(λ)+IL(λ)+G2(λ)
=G'(λ)+IL (λ) (2)
IL (λ) indicates the corresponding intergrade loss of wavelength X;Internal gain G ' (λ), external gain G (λ) and intergrade loss
It is linear relationship between IL (λ) three;
The gain control of EDFA is based on photodetector, and probe value is average value, average gain G external at this time are as follows:
Wherein, G and G ' respectively indicates external average gain, internal average gain, G1Indicate that first order amplifier averagely increases
Benefit, G2Indicate that second level amplifier average gain, IL indicate intergrade average loss;
When multi-wavelength amplifies, GR and GT is introduced, wherein GR indicates gain fluctuation (gain ripple), and GT indicates that gain is oblique
Rate (gain tilt);It is defined as shown in Fig. 2, mathematical formulae is as follows:
Gline(λi)=k*λi+c
GT=Gline(λn)-Gline(λ0)=k* (λn-λ0)
GR=max [G (λi)-Gline(λi)]-min[G(λi)-Gline(λi)] (4)
I=0....n
GlineIt is the linear fit function of gain and wavelength, k and c are the slope and intercept of formula (4) respectively;λiIndicate i-th
A wavelength;Start wavelength is λ0, termination wavelength is λn, unit nm;GT, GR unit are dB;
When VGEDFA amplifies for multi-wavelength, when G is set as different value, gain spectral is kept essentially constant, i.e. GT and GR
Meet index request (ideally GT, GR are remained unchanged);
For a VGEDFA, GT and G ' are linear,
GT=k2*G'+c2
=k2* (G-IL)+c2 (403)
Wherein k2 and c2 is respectively slope and intercept in formula (403), available by calibrating;K2 is negative value, C-band model
When enclosing, k2 is about -0.7~-0.9;
By formula (403) if it is found that G is remained unchanged in VGEDFA, GT and IL are linear;Otherwise IL is remained unchanged,
GT and G are linear;
In VGEDFA, if wavelength XiCorresponding internal gain G ' (λi) remain unchanged, when IL changes, when:
1)ΔIL(λi)=Δ IL, i.e. IL variable quantity and wavelength are unrelated, and the corresponding intergrade loss of any one wavelength becomes
Change amount is identical, by formula (2) it is found that G (λ at this timei) it is that overall variation and wavelength are unrelated, i.e., gain spectral shape does not become
Change, GR, GT are remained unchanged;ΔIL(λi) indicate wavelength XiCorresponding intergrade Dissipation change amount, Δ IL indicate that intergrade is average
Dissipation change amount;
2)ΔIL(λi) ≠ Δ IL, i.e. IL variable quantity is related to wavelength, the corresponding intergrade Dissipation change amount of different wave length
It is inconsistent, to Δ IL (λi) and λiIt carries out linear fit and obtains linear function Δ ILline(λi), calculate the corresponding Δ of Δ IL
Tilt is indicated with formula (401);By formula (2) it is found that gain spectral shape is changed in this case: GR, GT are no longer kept
It is constant;
ΔIL(λi)=IL (λi)-ILbg(λi)
ΔILline(λi)=k3* λi+c3 (401)
Δ Tilt=Δ ILline(λn)-ΔILline(λ0)=k3* (λn-λ0)
ILbgIt indicates intergrade background losses (background IL), ILbg(λi) indicate wavelength XiCorresponding intergrade sheet
Bottom loss;IL(λi) indicate wavelength XiCorresponding intergrade loss;K3, c3 are the slope and intercept of formula (401);Due to λn≥
λ0, therefore Δ Tilt and k3 positive negativity having the same;
Δ Tilt can be different value under different Δ IL, that is, correspond to different k3, c3 value;If Δ Tilt is the letter of Δ IL
Number is indicated with formula (405):
Δ Tilt=f (Δ ILline)=f (IL-ILbg) (405)
It is still linear relationship after the superposition of two lines formula, and slope and intercept are the relationships of superposition, therefore introduces Δ
After Tilt, GT and GTbaseRelationship it is as follows:
GT=GTbase+ΔTilt
ΔGT1=GT-GTbase=Δ Tilt (402)
ILbg=Gmax-G′base(404)
GmaxIndicate maxgain value, G 'baseIndicate benchmark internal gain;GTbaseIt is defined as reference gain slope, i.e.,
ILbg、Gmax、G’baseCorresponding gain slope;
Learn that G is remained unchanged by formula (403), the variation of IL causes GT to change, if variable quantity is Δ GT2:
ΔGT2=GT-GTbase=(G'-G 'base)*k2
=[(G-IL)-(G-ILbase)]*k2 (5)
=(ILbase-IL)*k2
ILbaseIndicate G, G 'baseCorresponding benchmark intergrade average loss, i.e. ILbase=G-G 'base;Formula (402) table
Bright, the introducing of Δ Tilt will lead to GR, GT variation, and wherein the variable quantity of GT is Δ GT1;Formula (5) indicates that the variation of IL can be led
GT under identical gain is caused to change, variable quantity is Δ GT2;By GT=GTbase+ΔGT1+ΔGT2It is found that keep GT constant,
As long as meeting formula (501);
It is further explained as, G when being some steady state value, the shadow as caused by Δ Tilt can be compensated by adjusting G ' or IL
It rings, when GT variable quantity meets formula (501), GT is remained unchanged;
ΔGT1+ΔGT2=0 (501)
Further abbreviation obtains:
It is learnt by formula (502), the variation tendency of IL (λ) can be designed as particular value, since K2 is negative value:
1) as Δ Tilt < 0, IL is greater than ILbase;
2) as Δ Tilt > 0, IL is less than ILbase;
When casacade multi-amplifier cascades, schematic diagram as shown in Figure 3, noise figure NF formula:
Unit in formula (7) is linear amount;
When amplifier is two-stage design, formula (7) abbreviation:
NF1Indicate first order amplifier noise index, NF2Indicate second level amplifier noise index;
When log unit is converted to linear unit, numerical value perseverance is greater than 0.Work as IL > ILbaseWhen, NF is learnt by formula (701)
(IL)<NF(ILbase), i.e., as Δ Tilt < 0, the case where G and GT are remained unchanged, since IL value increases, NF can obtain excellent
Change.Conversely, working as IL < ILbaseWhen, NF can be deteriorated.
When amplifier connection grade number expands to n (n > 2) grade from two-stage, above-mentioned conclusion is still set up;
The present invention is by designing specific wavelength XiIL (λ is lost in corresponding intergradei), the corresponding intergrade of different wave length
Dissipation change amount Δ IL (λi) different, as Δ Tilt < 0, with the reduction of gain, GT is remained unchanged, GR is without obvious deterioration
In the case where, noise figure NF can be optimized;
In order to which simulation calculation is convenient, two linear relationships are set:
Setting 1, formula (405) is set as linear relationship:
Δ Tilt=k1* (IL-ILbg) (6)
Formula (6) substitutes into formula (502), derives IL are as follows:
Set 2, IL (λi) with wavelength XiIt changes linearly, at this time IL=[IL (λ0)+IL(λn)]/2, IL (λi) formula can be used
(602) it is calculated:
It is described in detail below with a specific product, the VGEDFA parameter is as follows:
Wave-length coverage: 1529~1562nm
G:10~20dB
ILbg: 1dB defines IL for convenience of calculationbgIt is unrelated with wavelength.
Output power Output power:17dbm
It can be obtained by above-mentioned parameter:
λ0=1529nm, λn=1562nm, Δ λ=33nm.
Gmax=20dB, G 'base=21dB.
Using the optical design model of a routine, schematic diagram is as shown in Figure 4: wherein G1 indicates first order amplifier, G2
Indicate that second level amplifier, IL indicate variable intergrade loss.Pump laser is by optical splitter splitter pump power
It is divided into two parts, a part access first order amplifier G1, another part accesses second level amplifier G2;In simulation model
Splitter is 1:1 light splitting.
IL is set as fixed value by the first step, adjusts pump power, and emulation obtains one group of G and GT, and linear fit calculates
K2 value.IL is set as ILbg, simulation result is as shown in table 1.GT and G linear fit result can obtain k2=- as shown in figure (5)
0.74。
input(dBm) |
output(dBm) |
G(dB) |
GT(dB) |
Max GR(dB) |
Min GR(dB) |
Max NF(dB) |
‐3.00 |
15.45 |
18.45 |
1.25 |
0.20 |
‐0.23 |
4.66 |
‐3.00 |
16.12 |
19.12 |
0.75 |
0.15 |
‐0.15 |
4.59 |
‐3.00 |
17.00 |
20.00 |
0.10 |
0.12 |
‐0.10 |
4.52 |
‐3.00 |
17.72 |
20.72 |
‐0.43 |
0.13 |
‐0.15 |
4.46 |
‐3.00 |
18.64 |
21.64 |
‐1.11 |
0.21 |
‐0.24 |
4.40 |
Table 1
Second step, when k1=0,0.2, G=20,15,10 calculate IL according to formula (601), are calculated according to formula (6)
Δ Tilt out, calculated result are as follows.Wherein: not introducing Δ Tilt, i.e. IL=IL when k1=0base。
G1 |
G2 |
G3 |
20 |
15 |
10 |
ILbase1 |
ILbase2 |
ILbase3 |
‐1 |
‐6 |
‐11 |
IL1 |
IL2 |
IL3 |
‐1.00 |
‐4.94 |
‐8.87 |
ΔTilt1 |
ΔTilt2 |
ΔTilt3 |
0.00 |
‐0.79 |
‐1.57 |
Third step substitutes into IL obtained above, Δ Tilt value formula (602), calculates IL (λi), as schemed (6) institute
Show.
4th step, the above-mentioned IL (λ being calculatedi) simulation software is substituted into, emulate G=20,15,10 3 gain points.
The simulation result that gain spectral, NF are composed such as figure (7), figure (8), figure (9) are shown.
Simulation result arranges as shown in table 2.EDF (m) is Er-doped fiber (abbreviation erbium is fine) length, and unit is rice (m);It is very bright
Aobvious, when G=10dB, in the case that EDF length ratio is constant, when k1 is changed to 0.2 from 0, max NF can reduce about 0.9dB,
GT and GR are kept essentially constant at this time.K1=0 under theoretical case.
Table 2
Particularly, according in figure (9) NF spectrum it is found that long wavelength NF be greater than short long wave NF, it is long by adjusting two sections of erbium fibres
Degree ratio, Max NF can also be further decreased.For example erbium fibre length is changed to 8.5+9.2m from 8+9.7, Max NF is reduced
0.15dB, reference table 2.The adjustment of erbium fibre length ratio is that simplest thing, the present invention do not describe in detail in optical design.
By theoretical and simulation result it is found that designing different Δ Tilt can change under the premise of not sacrificing GR, GT
The corresponding noise figure of the small gain of VGEDFA.When Δ Tilt is less than 0, when small gain, can obtain lower noise figure.
From formula (3), formula (501) it is found why the Δ Tilt no matter introduced is worth, IL control method is all protected in VGEDFA
It holds constant.
In current VGEDFA, variable loss IL generally pass through VOA (Variable Optical attenuator) into
Row adjustment.For conventional VOA, Δ Tilt >=0, this VOA is unfavorable to the NF of small gain.Therefore one kind can be designed
VOA, with the reduction (attenuation is negative value) of its attenuation, Δ Tilt < 0, the NF of small gain can get optimization at this time.At present
Existing producer can design and make negative Δ Tilt VOA, and when VOA attenuation is -10dB, Δ Tilt can accomplish the left side -0.5dB
It is right.The VOA is had been introduced into actual product, NF can optimize 0.2dB or so under Min gain.
Δ Tilt can certainly be introduced by its device.Specific design belongs to devices field, and the present invention is not covered.
It should be noted last that the above specific embodiment is only used to illustrate the technical scheme of the present invention and not to limit it,
Although being described the invention in detail referring to example, those skilled in the art should understand that, it can be to the present invention
Technical solution be modified or replaced equivalently, without departing from the spirit and scope of the technical solution of the present invention, should all cover
In the scope of the claims of the present invention.