Attenuator: 50ohm/81dB 1dB step
Created@2012/12/15
finished by 2013/1/14, one month hard work
The basic tools for RF, you must have one.
Attenuator Plan and select box
2 box, the one given following attenuation step:
1dB 2dB 3dB 5dB
another :
10dB 20dB 20dB 20dB
Design goal:
*RL>60dB at DC
*RL>40dB at VHF
*usable to 70cm band
the die-casting aluminum box image: ( 9cmx3.6cmx3cm)
Create drill Stencil
drill with Stencil
when finished, the drill Stencil will use as secondary PCB box provide excellent shield.
checking after drill finished
.
Get Ideal resistor from 5% series
use %5 1/4W resistor, here is a method to get ideal resistor for attenuator: select a closed 5% resistor which big than the wanted value, then parallel another resistor to get perfect resistor. typically <1R to perfect attenuator value, even 0.5R.
First of all, you should have at least a 4.5 digital DVM or equivalent device.
for example, you want 5.769 ohm,
1) then pick 6.2R, in this step use a less value, ie, you 5% resistor had 6.0,6.1,6.2ohm, then use 6.0.
2) then try to parallel a 68ohm resistor on it, check it with you DVM, if you get ie, 5.3ohm, if you accept this, done. if not
3) 5.3ohm<5.769, we should try use large resistor, try 75R,82R ....
here is all the matched pair for your start to get your ideal resistor, after the "=" is the value i get, and I'm very happy to get these perfect value resistor.
perfect closed
1dB: shunt: 869.548 910||22k = 868 Input RL: >73dB
series: 5.769 6.2||68=5.75 Out RL:>73dB
perfect closed
2dB: shunt: 436.212 470||7.5k = 436.4 Input RL: >78dB
series: 11.615 12||430=11.60 Out RL: >78dB
perfect closed
3dB: shunt: 292.402 300||16k =292.5 Input RL: >75dB
series: 17.615 18||1.1k=17.6 Out RL: >75dB
perfect closed
5dB: shunt: 178.489 200||1.8k =178.6 Input RL: >60dB
series: 30.398 33||430= 30.49 Out RL: >60dB
-----------------------------------------------------------------------------------------------
perfect closed
10dB: shunt: 96.248 100||3.9k = 96.3 Input RL: >70dB
series: 71.151 75||1.3k= 71.1 Out RL: >70dB
perfect closed
20dB: shunt: 61.111 62||3.3k =61.10 Input RL: >70dB
series: 247.502 270(265.3)||3.9k(3.83)=247.8 Out RL: >70dB
Making secondary PCB shield box fit into the die-casting box
drill mask used as another PCB box's panel, then we need more PCB parts to form a box. the following image show how to make many same size parts one time.
When finished all parts, put them into die cast box, checking if the dimension is suitable.
Soldering the PCB shield box
NOTE:
1. use wire to connect bottom and the top layer of the PCB
2. use PCB strip to connect the switch for better performance at high frequncy. you should make through hole for inter connect.
3. use copper tape seal the bottom
DC checking
Use a 4 1/2 digital DVM, a 50R load, a battery to check the DC resistance and the DC attenuation. this ensure all of them connected properly and given a low Frequency attenuation reference.
battery -> attenuator -> 50R load
Attenuation=20*log(Vout/Vin)
1. resistance check
terminated by 50R, test another port resistance
1db/2db/3db/10db/20db : 50.05R
5dB: 50.10R
2. DC attenuate
1dB: in: 1.2983 out:1.1572 0.999 dB
2dB in:1.2978 out: 1.0316 1.994 dB
3dB in: 1.2976 out: 0.9186 3.000 dB
5dB in:1.2975 out: 0.7289 5.009 dB
10dB in:1.2980 out: 0.4101 10.007dB
20dB in:1.2977 out:0.12931 20.03 dB
@2003/9/22
few day ago i finished the 8307 Power Meter, which work really good up to 470Mhz(which i can test by a handy transceiver with 40dB pad attenuator) , so can verify the high frequency performance.
10Mhz testing(pending)
dam good.
144Mhz testing(pending)
really good.
455Mhz testing(pending)
1-10dB is fine, but with 20dB, only get 18~19db attenuation. refer following sections.
compensation the step attenuator up to 500Mhz
@2003/9/22
Here is the simulation with LTspice, a typical axial 1/4 Watt resistor might have 8nH parasitic inductance and 0.3pF pakage ((which should less than 8nH, i cut the lead is damn short). with these i do get the seems fit the reality performance.
Note: the lower blue line is the attenuation.
how to compensate this? learn something from the AD8307 compensate method, i can use a 3pF capacitor with long lead bend to a 5mm diameter inductor, witch should around 8nH~9nH. I got this simulate result:
Simulation tell me such things should work, let's check it tonight! so much effort to construct this device, but axial version seems hard to work to 500Mhz.
compensation Version
Final Result:
below the line is compassionate version, accuracy improved 2.x dB from 140M to 470Mhz.
(column 0 means the signal strength when enumerator set to 0, -10 represent the signal when 10 db switch in)
Conclusion @2013/9/23
*1- 10dB attenuation and are useful up to 500Mhz, less than 0.5dB error (up to 500Mhz).
*20dB attenuation are accuracy up to 150Mhz, error less than 0.5dB after compensation.
*from 200 to 450Mhz error less than -1dB.
*470Mhz get +1 error attenuation.
* 1-10dB + 2X20dB is useable, but 3x20dB won't work...... (might because signal leak?)
*Compensation is work but limited, seem, it's hard to tune it to cut at 900Mhz and keep error < 1dB