IC-R75 Modifications, AF Filter Analysis
In the page of IC-R75 Modifications, I have mentioned about AF filter design flaw.
I also recommended filter frequency change. Now, what will happen if I modify the filter circuit?
Here is the analysis of them.
There are two means to shift the cut-off frequency to higher position. One is capacitor replacement, and the other way is to replace resistors. Following analysis deals with both cases. In short, resistor replacement has disadvantage because it is affected more by source impedance.
BASE:
First of all, you see the ICOM designers intention. The two stage filter creates 24dB/oct (4 pole) Butterworth response if it is driven from low impedance source. The half-power point is 2.96kHz. The red line is the one.
CASE 0: ICOM Original
Now, the actual response of the original circuit is this. Source impedance of each FM, SSB, and AM has different value. So I calculated all of them. The line of AM//47k is the case with 47kohm attenuator as written in the diagram (I did not implement this attenuator, though).
Assumed 100 ohm at the analog switch.
The high source impedance deteriorate SSB and AM signal to a certain
extent.
CASE 1: C Replaced
Following graph is the case of capacitor replacement so that the cut-off becomes 4.4kHz (but not actually as you see). The schematic shows new capacitor values. Curves are proportional to case 0. Just shifted right by 1.5 times in frequency.
CASE 2: R Replaced
Here comes resistor replacement case. 90kohm resistors are added in parallel to the original 39kohm ones, making them to 27.2k.
As you see, the yellow curve tends to shift left, or becomes dull.
So, I recommend C-replacement (Case 1).Read on.
CASE 4: C Replaced & Rs Reduced
CASE 5: R Replaced & Rs Reduced
Yet other alternatives are these. They attempt to reduce the source impedance (Rs), followed by either C replaced filter or R replaced filter.
The blue 39kohm can be lower than that, say 22k or so to have better characteristic. However, 39k is convenient to have -3.6dB attenuation for AM and S-AM without additional resistors as red letters tell. Only two (39k and 68k) additions are needed if you accept -3.6dB for AMs. (Note: The actual loss by missing parallel resistor is 4.1dB, not 3.6. The 0.5dB margin is for signal alignment of all inputs.)
Curves show better fitness than cases 1 and 2.
This is the case 4 figure.
I implemented this circuit in my IC-R75 with no blue 47k for AM and S-AM.
And, this is the case 5.Case 4 is the best solution among them, followed by case 5 or 2. Of them case 5 with -3.6dB AM is the easiest rework because you do not remove chip components (but, ferroelectric caps introduce slight distortion).
[Additional Mod]
You need following modification additionally.
There is a simple RC filter at the power amp input. The cut-off of this filter is 3.4kHz, so I changed it to 7kHz.
See picture on the right.
[Reference]
If you plan to insert a separate filter rather than replacing R/C components, please consider this circuit. This is a five pole Chebyshev with 0.5dB ripple. This filter effectively cuts 5kHz beat caused by a near-by station. 564p is a 560p and 4p of stray cap. All filter components must be plus/minus 1%.
Use my Excel sheet on ==> this page for any change in filter parameters.
[Edit, 6/24/2014]
There was a voice in Yahoo forum re how my rework was. Here you see the detail. I regret my poor soldering though. Caps are glued togather to aviod vibration.
The patch includes AF gain reduction at M5282 as well as the filter frequency change and filter source impedance reduction. Refer to this page for those mods.
